From e0401f3f712353e53d936aeaaa8437aaea9a1c56 Mon Sep 17 00:00:00 2001
From: "dusan.milosavljevic@imgtec.com"
 <dusan.milosavljevic@imgtec.com@ce2b1a6d-e550-0410-aec6-3dcde31c8c00>
Date: Tue, 29 Jul 2014 18:02:26 +0000
Subject: [PATCH] MIPS64: Add support for architecture revision 6.

TEST=
BUG=
R=jkummerow@chromium.org, paul.lind@imgtec.com

Review URL: https://codereview.chromium.org/426863006

git-svn-id: https://v8.googlecode.com/svn/branches/bleeding_edge@22681 ce2b1a6d-e550-0410-aec6-3dcde31c8c00
---
 build/toolchain.gypi                 |  17 +-
 src/mips64/assembler-mips64.cc       | 514 ++++++++++++++++++++++++++-
 src/mips64/assembler-mips64.h        | 160 ++++++++-
 src/mips64/code-stubs-mips64.cc      |  31 +-
 src/mips64/constants-mips64.cc       |   4 +
 src/mips64/constants-mips64.h        | 102 +++++-
 src/mips64/disasm-mips64.cc          | 452 ++++++++++++++++++++---
 src/mips64/full-codegen-mips64.cc    |  19 +-
 src/mips64/lithium-codegen-mips64.cc |  48 +--
 src/mips64/lithium-mips64.cc         |   4 +-
 src/mips64/lithium-mips64.h          |   6 +-
 src/mips64/macro-assembler-mips64.cc | 360 +++++++++++++------
 src/mips64/macro-assembler-mips64.h  |   4 +
 src/mips64/simulator-mips64.cc       | 239 ++++++++++---
 test/cctest/test-api.cc              |   4 +-
 test/cctest/test-assembler-mips64.cc | 245 +++++++------
 test/cctest/test-disasm-mips64.cc    | 331 ++++++++++++-----
 test/mjsunit/mjsunit.status          |   3 +
 18 files changed, 2033 insertions(+), 510 deletions(-)

diff --git a/build/toolchain.gypi b/build/toolchain.gypi
index 929461351b..496cef0dbd 100644
--- a/build/toolchain.gypi
+++ b/build/toolchain.gypi
@@ -376,6 +376,14 @@
                     'cflags': ['-msoft-float'],
                     'ldflags': ['-msoft-float'],
                   }],
+                  ['mips_arch_variant=="r6"', {
+                    'cflags': ['-mips64r6', '-mabi=64', '-Wa,-mips64r6'],
+                    'ldflags': [
+                      '-mips64r6', '-mabi=64',
+                      '-Wl,--dynamic-linker=$(LDSO_PATH)',
+                      '-Wl,--rpath=$(LD_R_PATH)',
+                    ],
+                  }],
                   ['mips_arch_variant=="r2"', {
                     'cflags': ['-mips64r2', '-mabi=64', '-Wa,-mips64r2'],
                     'ldflags': [
@@ -384,9 +392,6 @@
                       '-Wl,--rpath=$(LD_R_PATH)',
                     ],
                   }],
-                  ['mips_arch_variant=="loongson"', {
-                    'cflags': ['-mips3', '-Wa,-mips3'],
-                  }],
                 ],
               }],
             ],
@@ -406,12 +411,12 @@
               '__mips_soft_float=1'
             ],
           }],
+          ['mips_arch_variant=="r6"', {
+            'defines': ['_MIPS_ARCH_MIPS64R6',],
+          }],
           ['mips_arch_variant=="r2"', {
             'defines': ['_MIPS_ARCH_MIPS64R2',],
           }],
-          ['mips_arch_variant=="loongson"', {
-            'defines': ['_MIPS_ARCH_LOONGSON',],
-          }],
         ],
       }],  # v8_target_arch=="mips64el"
       ['v8_target_arch=="x64"', {
diff --git a/src/mips64/assembler-mips64.cc b/src/mips64/assembler-mips64.cc
index 2796c95b37..48cb90a0e7 100644
--- a/src/mips64/assembler-mips64.cc
+++ b/src/mips64/assembler-mips64.cc
@@ -485,7 +485,9 @@ bool Assembler::IsBranch(Instr instr) {
       opcode == BGTZL ||
       (opcode == REGIMM && (rt_field == BLTZ || rt_field == BGEZ ||
                             rt_field == BLTZAL || rt_field == BGEZAL)) ||
-      (opcode == COP1 && rs_field == BC1);  // Coprocessor branch.
+      (opcode == COP1 && rs_field == BC1) ||  // Coprocessor branch.
+      (opcode == COP1 && rs_field == BC1EQZ) ||
+      (opcode == COP1 && rs_field == BC1NEZ);
 }
 
 
@@ -969,7 +971,6 @@ void Assembler::GenInstrJump(Opcode opcode,
 // Returns the next free trampoline entry.
 int32_t Assembler::get_trampoline_entry(int32_t pos) {
   int32_t trampoline_entry = kInvalidSlotPos;
-
   if (!internal_trampoline_exception_) {
     if (trampoline_.start() > pos) {
      trampoline_entry = trampoline_.take_slot();
@@ -985,7 +986,6 @@ int32_t Assembler::get_trampoline_entry(int32_t pos) {
 
 uint64_t Assembler::jump_address(Label* L) {
   int64_t target_pos;
-
   if (L->is_bound()) {
     target_pos = L->pos();
   } else {
@@ -1007,7 +1007,6 @@ uint64_t Assembler::jump_address(Label* L) {
 
 int32_t Assembler::branch_offset(Label* L, bool jump_elimination_allowed) {
   int32_t target_pos;
-
   if (L->is_bound()) {
     target_pos = L->pos();
   } else {
@@ -1032,6 +1031,86 @@ int32_t Assembler::branch_offset(Label* L, bool jump_elimination_allowed) {
 }
 
 
+int32_t Assembler::branch_offset_compact(Label* L,
+    bool jump_elimination_allowed) {
+  int32_t target_pos;
+  if (L->is_bound()) {
+    target_pos = L->pos();
+  } else {
+    if (L->is_linked()) {
+      target_pos = L->pos();
+      L->link_to(pc_offset());
+    } else {
+      L->link_to(pc_offset());
+      if (!trampoline_emitted_) {
+        unbound_labels_count_++;
+        next_buffer_check_ -= kTrampolineSlotsSize;
+      }
+      return kEndOfChain;
+    }
+  }
+
+  int32_t offset = target_pos - pc_offset();
+  ASSERT((offset & 3) == 0);
+  ASSERT(is_int16(offset >> 2));
+
+  return offset;
+}
+
+
+int32_t Assembler::branch_offset21(Label* L, bool jump_elimination_allowed) {
+  int32_t target_pos;
+  if (L->is_bound()) {
+    target_pos = L->pos();
+  } else {
+    if (L->is_linked()) {
+      target_pos = L->pos();
+      L->link_to(pc_offset());
+    } else {
+      L->link_to(pc_offset());
+      if (!trampoline_emitted_) {
+        unbound_labels_count_++;
+        next_buffer_check_ -= kTrampolineSlotsSize;
+      }
+      return kEndOfChain;
+    }
+  }
+
+  int32_t offset = target_pos - (pc_offset() + kBranchPCOffset);
+  ASSERT((offset & 3) == 0);
+  ASSERT(((offset >> 2) & 0xFFE00000) == 0);  // Offset is 21bit width.
+
+  return offset;
+}
+
+
+int32_t Assembler::branch_offset21_compact(Label* L,
+    bool jump_elimination_allowed) {
+  int32_t target_pos;
+  if (L->is_bound()) {
+    target_pos = L->pos();
+  } else {
+    if (L->is_linked()) {
+      target_pos = L->pos();
+      L->link_to(pc_offset());
+    } else {
+      L->link_to(pc_offset());
+      if (!trampoline_emitted_) {
+        unbound_labels_count_++;
+        next_buffer_check_ -= kTrampolineSlotsSize;
+      }
+      return kEndOfChain;
+    }
+  }
+
+  int32_t offset = target_pos - pc_offset();
+  ASSERT((offset & 3) == 0);
+  ASSERT(((offset >> 2) & 0xFFE00000) == 0);  // Offset is 21bit width.
+
+  return offset;
+}
+
+
 void Assembler::label_at_put(Label* L, int at_offset) {
   int target_pos;
   if (L->is_bound()) {
@@ -1085,7 +1164,33 @@ void Assembler::bgez(Register rs, int16_t offset) {
 }
 
 
+void Assembler::bgezc(Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rt.is(zero_reg)));
+  GenInstrImmediate(BLEZL, rt, rt, offset);
+}
+
+
+void Assembler::bgeuc(Register rs, Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rs.is(zero_reg)));
+  ASSERT(!(rt.is(zero_reg)));
+  ASSERT(rs.code() != rt.code());
+  GenInstrImmediate(BLEZ, rs, rt, offset);
+}
+
+
+void Assembler::bgec(Register rs, Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rs.is(zero_reg)));
+  ASSERT(!(rt.is(zero_reg)));
+  ASSERT(rs.code() != rt.code());
+  GenInstrImmediate(BLEZL, rs, rt, offset);
+}
+
+
 void Assembler::bgezal(Register rs, int16_t offset) {
+  ASSERT(kArchVariant != kMips64r6 || rs.is(zero_reg));
   BlockTrampolinePoolScope block_trampoline_pool(this);
   positions_recorder()->WriteRecordedPositions();
   GenInstrImmediate(REGIMM, rs, BGEZAL, offset);
@@ -1100,6 +1205,13 @@ void Assembler::bgtz(Register rs, int16_t offset) {
 }
 
 
+void Assembler::bgtzc(Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rt.is(zero_reg)));
+  GenInstrImmediate(BGTZL, zero_reg, rt, offset);
+}
+
+
 void Assembler::blez(Register rs, int16_t offset) {
   BlockTrampolinePoolScope block_trampoline_pool(this);
   GenInstrImmediate(BLEZ, rs, zero_reg, offset);
@@ -1107,6 +1219,38 @@ void Assembler::blez(Register rs, int16_t offset) {
 }
 
 
+void Assembler::blezc(Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rt.is(zero_reg)));
+  GenInstrImmediate(BLEZL, zero_reg, rt, offset);
+}
+
+
+void Assembler::bltzc(Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rt.is(zero_reg)));
+  GenInstrImmediate(BGTZL, rt, rt, offset);
+}
+
+
+void Assembler::bltuc(Register rs, Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rs.is(zero_reg)));
+  ASSERT(!(rt.is(zero_reg)));
+  ASSERT(rs.code() != rt.code());
+  GenInstrImmediate(BGTZ, rs, rt, offset);
+}
+
+
+void Assembler::bltc(Register rs, Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rs.is(zero_reg)));
+  ASSERT(!(rt.is(zero_reg)));
+  ASSERT(rs.code() != rt.code());
+  GenInstrImmediate(BGTZL, rs, rt, offset);
+}
+
+
 void Assembler::bltz(Register rs, int16_t offset) {
   BlockTrampolinePoolScope block_trampoline_pool(this);
   GenInstrImmediate(REGIMM, rs, BLTZ, offset);
@@ -1115,6 +1259,7 @@ void Assembler::bltz(Register rs, int16_t offset) {
 
 
 void Assembler::bltzal(Register rs, int16_t offset) {
+  ASSERT(kArchVariant != kMips64r6 || rs.is(zero_reg));
   BlockTrampolinePoolScope block_trampoline_pool(this);
   positions_recorder()->WriteRecordedPositions();
   GenInstrImmediate(REGIMM, rs, BLTZAL, offset);
@@ -1129,6 +1274,101 @@ void Assembler::bne(Register rs, Register rt, int16_t offset) {
 }
 
 
+void Assembler::bovc(Register rs, Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rs.is(zero_reg)));
+  ASSERT(rs.code() >= rt.code());
+  GenInstrImmediate(ADDI, rs, rt, offset);
+}
+
+
+void Assembler::bnvc(Register rs, Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rs.is(zero_reg)));
+  ASSERT(rs.code() >= rt.code());
+  GenInstrImmediate(DADDI, rs, rt, offset);
+}
+
+
+void Assembler::blezalc(Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rt.is(zero_reg)));
+  GenInstrImmediate(BLEZ, zero_reg, rt, offset);
+}
+
+
+void Assembler::bgezalc(Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rt.is(zero_reg)));
+  GenInstrImmediate(BLEZ, rt, rt, offset);
+}
+
+
+void Assembler::bgezall(Register rs, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rs.is(zero_reg)));
+  GenInstrImmediate(REGIMM, rs, BGEZALL, offset);
+}
+
+
+void Assembler::bltzalc(Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rt.is(zero_reg)));
+  GenInstrImmediate(BGTZ, rt, rt, offset);
+}
+
+
+void Assembler::bgtzalc(Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rt.is(zero_reg)));
+  GenInstrImmediate(BGTZ, zero_reg, rt, offset);
+}
+
+
+void Assembler::beqzalc(Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rt.is(zero_reg)));
+  GenInstrImmediate(ADDI, zero_reg, rt, offset);
+}
+
+
+void Assembler::bnezalc(Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rt.is(zero_reg)));
+  GenInstrImmediate(DADDI, zero_reg, rt, offset);
+}
+
+
+void Assembler::beqc(Register rs, Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(rs.code() < rt.code());
+  GenInstrImmediate(ADDI, rs, rt, offset);
+}
+
+
+void Assembler::beqzc(Register rs, int32_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rs.is(zero_reg)));
+  Instr instr = BEQZC | (rs.code() << kRsShift) | offset;
+  emit(instr);
+}
+
+
+void Assembler::bnec(Register rs, Register rt, int16_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(rs.code() < rt.code());
+  GenInstrImmediate(DADDI, rs, rt, offset);
+}
+
+
+void Assembler::bnezc(Register rs, int32_t offset) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(!(rs.is(zero_reg)));
+  Instr instr = BNEZC | (rs.code() << kRsShift) | offset;
+  emit(instr);
+}
+
+
 void Assembler::j(int64_t target) {
 #if DEBUG
   // Get pc of delay slot.
@@ -1142,12 +1382,16 @@ void Assembler::j(int64_t target) {
 
 
 void Assembler::jr(Register rs) {
-  BlockTrampolinePoolScope block_trampoline_pool(this);
-  if (rs.is(ra)) {
-    positions_recorder()->WriteRecordedPositions();
+  if (kArchVariant != kMips64r6) {
+    BlockTrampolinePoolScope block_trampoline_pool(this);
+    if (rs.is(ra)) {
+      positions_recorder()->WriteRecordedPositions();
+    }
+    GenInstrRegister(SPECIAL, rs, zero_reg, zero_reg, 0, JR);
+    BlockTrampolinePoolFor(1);  // For associated delay slot.
+  } else {
+    jalr(rs, zero_reg);
   }
-  GenInstrRegister(SPECIAL, rs, zero_reg, zero_reg, 0, JR);
-  BlockTrampolinePoolFor(1);  // For associated delay slot.
 }
 
 
@@ -1218,16 +1462,64 @@ void Assembler::subu(Register rd, Register rs, Register rt) {
 
 
 void Assembler::mul(Register rd, Register rs, Register rt) {
-  GenInstrRegister(SPECIAL2, rs, rt, rd, 0, MUL);
+  if (kArchVariant == kMips64r6) {
+      GenInstrRegister(SPECIAL, rs, rt, rd, MUL_OP, MUL_MUH);
+  } else {
+      GenInstrRegister(SPECIAL2, rs, rt, rd, 0, MUL);
+  }
+}
+
+
+void Assembler::muh(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MUH_OP, MUL_MUH);
+}
+
+
+void Assembler::mulu(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MUL_OP, MUL_MUH_U);
+}
+
+
+void Assembler::muhu(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MUH_OP, MUL_MUH_U);
+}
+
+
+void Assembler::dmul(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MUL_OP, D_MUL_MUH);
+}
+
+
+void Assembler::dmuh(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MUH_OP, D_MUL_MUH);
+}
+
+
+void Assembler::dmulu(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MUL_OP, D_MUL_MUH_U);
+}
+
+
+void Assembler::dmuhu(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MUH_OP, D_MUL_MUH_U);
 }
 
 
 void Assembler::mult(Register rs, Register rt) {
+  ASSERT(kArchVariant != kMips64r6);
   GenInstrRegister(SPECIAL, rs, rt, zero_reg, 0, MULT);
 }
 
 
 void Assembler::multu(Register rs, Register rt) {
+  ASSERT(kArchVariant != kMips64r6);
   GenInstrRegister(SPECIAL, rs, rt, zero_reg, 0, MULTU);
 }
 
@@ -1242,11 +1534,35 @@ void Assembler::div(Register rs, Register rt) {
 }
 
 
+void Assembler::div(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, DIV_OP, DIV_MOD);
+}
+
+
+void Assembler::mod(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MOD_OP, DIV_MOD);
+}
+
+
 void Assembler::divu(Register rs, Register rt) {
   GenInstrRegister(SPECIAL, rs, rt, zero_reg, 0, DIVU);
 }
 
 
+void Assembler::divu(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, DIV_OP, DIV_MOD_U);
+}
+
+
+void Assembler::modu(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MOD_OP, DIV_MOD_U);
+}
+
+
 void Assembler::daddu(Register rd, Register rs, Register rt) {
   GenInstrRegister(SPECIAL, rs, rt, rd, 0, DADDU);
 }
@@ -1272,11 +1588,35 @@ void Assembler::ddiv(Register rs, Register rt) {
 }
 
 
+void Assembler::ddiv(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, DIV_OP, D_DIV_MOD);
+}
+
+
+void Assembler::dmod(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MOD_OP, D_DIV_MOD);
+}
+
+
 void Assembler::ddivu(Register rs, Register rt) {
   GenInstrRegister(SPECIAL, rs, rt, zero_reg, 0, DDIVU);
 }
 
 
+void Assembler::ddivu(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, DIV_OP, D_DIV_MOD_U);
+}
+
+
+void Assembler::dmodu(Register rd, Register rs, Register rt) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, MOD_OP, D_DIV_MOD_U);
+}
+
+
 // Logical.
 
 void Assembler::and_(Register rd, Register rs, Register rt) {
@@ -1566,6 +1906,32 @@ void Assembler::lui(Register rd, int32_t j) {
 }
 
 
+void Assembler::aui(Register rs, Register rt, int32_t j) {
+  // This instruction uses same opcode as 'lui'. The difference in encoding is
+  // 'lui' has zero reg. for rs field.
+  ASSERT(is_uint16(j));
+  GenInstrImmediate(LUI, rs, rt, j);
+}
+
+
+void Assembler::daui(Register rs, Register rt, int32_t j) {
+  ASSERT(is_uint16(j));
+  GenInstrImmediate(DAUI, rs, rt, j);
+}
+
+
+void Assembler::dahi(Register rs, int32_t j) {
+  ASSERT(is_uint16(j));
+  GenInstrImmediate(REGIMM, rs, DAHI, j);
+}
+
+
+void Assembler::dati(Register rs, int32_t j) {
+  ASSERT(is_uint16(j));
+  GenInstrImmediate(REGIMM, rs, DATI, j);
+}
+
+
 void Assembler::ldl(Register rd, const MemOperand& rs) {
   GenInstrImmediate(LDL, rs.rm(), rd, rs.offset_);
 }
@@ -1747,17 +2113,73 @@ void Assembler::movf(Register rd, Register rs, uint16_t cc) {
 }
 
 
+void Assembler::sel(SecondaryField fmt, FPURegister fd,
+    FPURegister ft, FPURegister fs, uint8_t sel) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(fmt == D);
+  ASSERT(fmt == S);
+
+  Instr instr = COP1 | fmt << kRsShift | ft.code() << kFtShift |
+      fs.code() << kFsShift | fd.code() << kFdShift | SEL;
+  emit(instr);
+}
+
+
+// GPR.
+void Assembler::seleqz(Register rs, Register rt, Register rd) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, 0, SELEQZ_S);
+}
+
+
+// FPR.
+void Assembler::seleqz(SecondaryField fmt, FPURegister fd,
+    FPURegister ft, FPURegister fs) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(fmt == D);
+  ASSERT(fmt == S);
+
+  Instr instr = COP1 | fmt << kRsShift | ft.code() << kFtShift |
+      fs.code() << kFsShift | fd.code() << kFdShift | SELEQZ_C;
+  emit(instr);
+}
+
+
+// GPR.
+void Assembler::selnez(Register rs, Register rt, Register rd) {
+  ASSERT(kArchVariant == kMips64r6);
+  GenInstrRegister(SPECIAL, rs, rt, rd, 0, SELNEZ_S);
+}
+
+
+// FPR.
+void Assembler::selnez(SecondaryField fmt, FPURegister fd,
+    FPURegister ft, FPURegister fs) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT(fmt == D);
+  ASSERT(fmt == S);
+
+  Instr instr = COP1 | fmt << kRsShift | ft.code() << kFtShift |
+      fs.code() << kFsShift | fd.code() << kFdShift | SELNEZ_C;
+  emit(instr);
+}
+
+
 // Bit twiddling.
 void Assembler::clz(Register rd, Register rs) {
-  // Clz instr requires same GPR number in 'rd' and 'rt' fields.
-  GenInstrRegister(SPECIAL2, rs, rd, rd, 0, CLZ);
+  if (kArchVariant != kMips64r6) {
+    // Clz instr requires same GPR number in 'rd' and 'rt' fields.
+    GenInstrRegister(SPECIAL2, rs, rd, rd, 0, CLZ);
+  } else {
+    GenInstrRegister(SPECIAL, rs, zero_reg, rd, 1, CLZ_R6);
+  }
 }
 
 
 void Assembler::ins_(Register rt, Register rs, uint16_t pos, uint16_t size) {
   // Should be called via MacroAssembler::Ins.
   // Ins instr has 'rt' field as dest, and two uint5: msb, lsb.
-  ASSERT(kArchVariant == kMips64r2);
+  ASSERT((kArchVariant == kMips64r2) || (kArchVariant == kMips64r6));
   GenInstrRegister(SPECIAL3, rs, rt, pos + size - 1, pos, INS);
 }
 
@@ -1765,13 +2187,12 @@ void Assembler::ins_(Register rt, Register rs, uint16_t pos, uint16_t size) {
 void Assembler::ext_(Register rt, Register rs, uint16_t pos, uint16_t size) {
   // Should be called via MacroAssembler::Ext.
   // Ext instr has 'rt' field as dest, and two uint5: msb, lsb.
-  ASSERT(kArchVariant == kMips64r2);
+  ASSERT(kArchVariant == kMips64r2 || kArchVariant == kMips64r6);
   GenInstrRegister(SPECIAL3, rs, rt, size - 1, pos, EXT);
 }
 
 
 void Assembler::pref(int32_t hint, const MemOperand& rs) {
-  ASSERT(kArchVariant != kLoongson);
   ASSERT(is_uint5(hint) && is_uint16(rs.offset_));
   Instr instr = PREF | (rs.rm().code() << kRsShift) | (hint << kRtShift)
       | (rs.offset_);
@@ -1870,7 +2291,6 @@ void Assembler::mul_d(FPURegister fd, FPURegister fs, FPURegister ft) {
 
 void Assembler::madd_d(FPURegister fd, FPURegister fr, FPURegister fs,
     FPURegister ft) {
-  ASSERT(kArchVariant != kLoongson);
   GenInstrRegister(COP1X, fr, ft, fs, fd, MADD_D);
 }
 
@@ -2006,6 +2426,38 @@ void Assembler::ceil_l_d(FPURegister fd, FPURegister fs) {
 }
 
 
+void Assembler::min(SecondaryField fmt, FPURegister fd, FPURegister ft,
+    FPURegister fs) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT((fmt == D) || (fmt == S));
+  GenInstrRegister(COP1, fmt, ft, fs, fd, MIN);
+}
+
+
+void Assembler::mina(SecondaryField fmt, FPURegister fd, FPURegister ft,
+    FPURegister fs) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT((fmt == D) || (fmt == S));
+  GenInstrRegister(COP1, fmt, ft, fs, fd, MINA);
+}
+
+
+void Assembler::max(SecondaryField fmt, FPURegister fd, FPURegister ft,
+    FPURegister fs) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT((fmt == D) || (fmt == S));
+  GenInstrRegister(COP1, fmt, ft, fs, fd, MAX);
+}
+
+
+void Assembler::maxa(SecondaryField fmt, FPURegister fd, FPURegister ft,
+    FPURegister fs) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT((fmt == D) || (fmt == S));
+  GenInstrRegister(COP1, fmt, ft, fs, fd, MAXA);
+}
+
+
 void Assembler::cvt_s_w(FPURegister fd, FPURegister fs) {
   GenInstrRegister(COP1, W, f0, fs, fd, CVT_S_W);
 }
@@ -2038,12 +2490,38 @@ void Assembler::cvt_d_s(FPURegister fd, FPURegister fs) {
 }
 
 
-// Conditions.
+// Conditions for >= MIPSr6.
+void Assembler::cmp(FPUCondition cond, SecondaryField fmt,
+    FPURegister fd, FPURegister fs, FPURegister ft) {
+  ASSERT(kArchVariant == kMips64r6);
+  ASSERT((fmt & ~(31 << kRsShift)) == 0);
+  Instr instr = COP1 | fmt | ft.code() << kFtShift |
+      fs.code() << kFsShift | fd.code() << kFdShift | (0 << 5) | cond;
+  emit(instr);
+}
+
+
+void Assembler::bc1eqz(int16_t offset, FPURegister ft) {
+  ASSERT(kArchVariant == kMips64r6);
+  Instr instr = COP1 | BC1EQZ | ft.code() << kFtShift | (offset & kImm16Mask);
+  emit(instr);
+}
+
+
+void Assembler::bc1nez(int16_t offset, FPURegister ft) {
+  ASSERT(kArchVariant == kMips64r6);
+  Instr instr = COP1 | BC1NEZ | ft.code() << kFtShift | (offset & kImm16Mask);
+  emit(instr);
+}
+
+
+// Conditions for < MIPSr6.
 void Assembler::c(FPUCondition cond, SecondaryField fmt,
     FPURegister fs, FPURegister ft, uint16_t cc) {
+  ASSERT(kArchVariant != kMips64r6);
   ASSERT(is_uint3(cc));
   ASSERT((fmt & ~(31 << kRsShift)) == 0);
-  Instr instr = COP1 | fmt | ft.code() << 16 | fs.code() << kFsShift
+  Instr instr = COP1 | fmt | ft.code() << kFtShift | fs.code() << kFsShift
       | cc << 8 | 3 << 4 | cond;
   emit(instr);
 }
diff --git a/src/mips64/assembler-mips64.h b/src/mips64/assembler-mips64.h
index 395ab77d8a..2345bf5afd 100644
--- a/src/mips64/assembler-mips64.h
+++ b/src/mips64/assembler-mips64.h
@@ -457,11 +457,20 @@ class Assembler : public AssemblerBase {
   // position. Links the label to the current position if it is still unbound.
   // Manages the jump elimination optimization if the second parameter is true.
   int32_t branch_offset(Label* L, bool jump_elimination_allowed);
+  int32_t branch_offset_compact(Label* L, bool jump_elimination_allowed);
+  int32_t branch_offset21(Label* L, bool jump_elimination_allowed);
+  int32_t branch_offset21_compact(Label* L, bool jump_elimination_allowed);
   int32_t shifted_branch_offset(Label* L, bool jump_elimination_allowed) {
     int32_t o = branch_offset(L, jump_elimination_allowed);
     ASSERT((o & 3) == 0);   // Assert the offset is aligned.
     return o >> 2;
   }
+  int32_t shifted_branch_offset_compact(Label* L,
+      bool jump_elimination_allowed) {
+    int32_t o = branch_offset_compact(L, jump_elimination_allowed);
+    ASSERT((o & 3) == 0);   // Assert the offset is aligned.
+    return o >> 2;
+  }
   uint64_t jump_address(Label* L);
 
   // Puts a labels target address at the given position.
@@ -617,15 +626,100 @@ class Assembler : public AssemblerBase {
     beq(rs, rt, branch_offset(L, false) >> 2);
   }
   void bgez(Register rs, int16_t offset);
+  void bgezc(Register rt, int16_t offset);
+  void bgezc(Register rt, Label* L) {
+    bgezc(rt, branch_offset_compact(L, false)>>2);
+  }
+  void bgeuc(Register rs, Register rt, int16_t offset);
+  void bgeuc(Register rs, Register rt, Label* L) {
+    bgeuc(rs, rt, branch_offset_compact(L, false)>>2);
+  }
+  void bgec(Register rs, Register rt, int16_t offset);
+  void bgec(Register rs, Register rt, Label* L) {
+    bgec(rs, rt, branch_offset_compact(L, false)>>2);
+  }
   void bgezal(Register rs, int16_t offset);
+  void bgezalc(Register rt, int16_t offset);
+  void bgezalc(Register rt, Label* L) {
+    bgezalc(rt, branch_offset_compact(L, false)>>2);
+  }
+  void bgezall(Register rs, int16_t offset);
+  void bgezall(Register rs, Label* L) {
+    bgezall(rs, branch_offset(L, false)>>2);
+  }
   void bgtz(Register rs, int16_t offset);
+  void bgtzc(Register rt, int16_t offset);
+  void bgtzc(Register rt, Label* L) {
+    bgtzc(rt, branch_offset_compact(L, false)>>2);
+  }
   void blez(Register rs, int16_t offset);
+  void blezc(Register rt, int16_t offset);
+  void blezc(Register rt, Label* L) {
+    blezc(rt, branch_offset_compact(L, false)>>2);
+  }
   void bltz(Register rs, int16_t offset);
+  void bltzc(Register rt, int16_t offset);
+  void bltzc(Register rt, Label* L) {
+    bltzc(rt, branch_offset_compact(L, false)>>2);
+  }
+  void bltuc(Register rs, Register rt, int16_t offset);
+  void bltuc(Register rs, Register rt, Label* L) {
+    bltuc(rs, rt, branch_offset_compact(L, false)>>2);
+  }
+  void bltc(Register rs, Register rt, int16_t offset);
+  void bltc(Register rs, Register rt, Label* L) {
+    bltc(rs, rt, branch_offset_compact(L, false)>>2);
+  }
+
   void bltzal(Register rs, int16_t offset);
+  void blezalc(Register rt, int16_t offset);
+  void blezalc(Register rt, Label* L) {
+    blezalc(rt, branch_offset_compact(L, false)>>2);
+  }
+  void bltzalc(Register rt, int16_t offset);
+  void bltzalc(Register rt, Label* L) {
+    bltzalc(rt, branch_offset_compact(L, false)>>2);
+  }
+  void bgtzalc(Register rt, int16_t offset);
+  void bgtzalc(Register rt, Label* L) {
+    bgtzalc(rt, branch_offset_compact(L, false)>>2);
+  }
+  void beqzalc(Register rt, int16_t offset);
+  void beqzalc(Register rt, Label* L) {
+    beqzalc(rt, branch_offset_compact(L, false)>>2);
+  }
+  void beqc(Register rs, Register rt, int16_t offset);
+  void beqc(Register rs, Register rt, Label* L) {
+    beqc(rs, rt, branch_offset_compact(L, false)>>2);
+  }
+  void beqzc(Register rs, int32_t offset);
+  void beqzc(Register rs, Label* L) {
+    beqzc(rs, branch_offset21_compact(L, false)>>2);
+  }
+  void bnezalc(Register rt, int16_t offset);
+  void bnezalc(Register rt, Label* L) {
+    bnezalc(rt, branch_offset_compact(L, false)>>2);
+  }
+  void bnec(Register rs, Register rt, int16_t offset);
+  void bnec(Register rs, Register rt, Label* L) {
+    bnec(rs, rt, branch_offset_compact(L, false)>>2);
+  }
+  void bnezc(Register rt, int32_t offset);
+  void bnezc(Register rt, Label* L) {
+    bnezc(rt, branch_offset21_compact(L, false)>>2);
+  }
   void bne(Register rs, Register rt, int16_t offset);
   void bne(Register rs, Register rt, Label* L) {
     bne(rs, rt, branch_offset(L, false)>>2);
   }
+  void bovc(Register rs, Register rt, int16_t offset);
+  void bovc(Register rs, Register rt, Label* L) {
+    bovc(rs, rt, branch_offset_compact(L, false)>>2);
+  }
+  void bnvc(Register rs, Register rt, int16_t offset);
+  void bnvc(Register rs, Register rt, Label* L) {
+    bnvc(rs, rt, branch_offset_compact(L, false)>>2);
+  }
 
   // Never use the int16_t b(l)cond version with a branch offset
   // instead of using the Label* version.
@@ -644,17 +738,34 @@ class Assembler : public AssemblerBase {
   // Arithmetic.
   void addu(Register rd, Register rs, Register rt);
   void subu(Register rd, Register rs, Register rt);
-  void mult(Register rs, Register rt);
-  void multu(Register rs, Register rt);
+
   void div(Register rs, Register rt);
   void divu(Register rs, Register rt);
+  void ddiv(Register rs, Register rt);
+  void ddivu(Register rs, Register rt);
+  void div(Register rd, Register rs, Register rt);
+  void divu(Register rd, Register rs, Register rt);
+  void ddiv(Register rd, Register rs, Register rt);
+  void ddivu(Register rd, Register rs, Register rt);
+  void mod(Register rd, Register rs, Register rt);
+  void modu(Register rd, Register rs, Register rt);
+  void dmod(Register rd, Register rs, Register rt);
+  void dmodu(Register rd, Register rs, Register rt);
+
   void mul(Register rd, Register rs, Register rt);
+  void muh(Register rd, Register rs, Register rt);
+  void mulu(Register rd, Register rs, Register rt);
+  void muhu(Register rd, Register rs, Register rt);
+  void mult(Register rs, Register rt);
+  void multu(Register rs, Register rt);
+  void dmul(Register rd, Register rs, Register rt);
+  void dmuh(Register rd, Register rs, Register rt);
+  void dmulu(Register rd, Register rs, Register rt);
+  void dmuhu(Register rd, Register rs, Register rt);
   void daddu(Register rd, Register rs, Register rt);
   void dsubu(Register rd, Register rs, Register rt);
   void dmult(Register rs, Register rt);
   void dmultu(Register rs, Register rt);
-  void ddiv(Register rs, Register rt);
-  void ddivu(Register rs, Register rt);
 
   void addiu(Register rd, Register rs, int32_t j);
   void daddiu(Register rd, Register rs, int32_t j);
@@ -669,6 +780,10 @@ class Assembler : public AssemblerBase {
   void ori(Register rd, Register rs, int32_t j);
   void xori(Register rd, Register rs, int32_t j);
   void lui(Register rd, int32_t j);
+  void aui(Register rs, Register rt, int32_t j);
+  void daui(Register rs, Register rt, int32_t j);
+  void dahi(Register rs, int32_t j);
+  void dati(Register rs, int32_t j);
 
   // Shifts.
   // Please note: sll(zero_reg, zero_reg, x) instructions are reserved as nop
@@ -751,6 +866,15 @@ class Assembler : public AssemblerBase {
   void movt(Register rd, Register rs, uint16_t cc = 0);
   void movf(Register rd, Register rs, uint16_t cc = 0);
 
+  void sel(SecondaryField fmt, FPURegister fd, FPURegister ft,
+      FPURegister fs, uint8_t sel);
+  void seleqz(Register rs, Register rt, Register rd);
+  void seleqz(SecondaryField fmt, FPURegister fd, FPURegister ft,
+      FPURegister fs);
+  void selnez(Register rs, Register rt, Register rd);
+  void selnez(SecondaryField fmt, FPURegister fd, FPURegister ft,
+      FPURegister fs);
+
   // Bit twiddling.
   void clz(Register rd, Register rs);
   void ins_(Register rt, Register rs, uint16_t pos, uint16_t size);
@@ -810,6 +934,11 @@ class Assembler : public AssemblerBase {
   void ceil_l_s(FPURegister fd, FPURegister fs);
   void ceil_l_d(FPURegister fd, FPURegister fs);
 
+  void min(SecondaryField fmt, FPURegister fd, FPURegister ft, FPURegister fs);
+  void mina(SecondaryField fmt, FPURegister fd, FPURegister ft, FPURegister fs);
+  void max(SecondaryField fmt, FPURegister fd, FPURegister ft, FPURegister fs);
+  void maxa(SecondaryField fmt, FPURegister fd, FPURegister ft, FPURegister fs);
+
   void cvt_s_w(FPURegister fd, FPURegister fs);
   void cvt_s_l(FPURegister fd, FPURegister fs);
   void cvt_s_d(FPURegister fd, FPURegister fs);
@@ -818,14 +947,31 @@ class Assembler : public AssemblerBase {
   void cvt_d_l(FPURegister fd, FPURegister fs);
   void cvt_d_s(FPURegister fd, FPURegister fs);
 
-  // Conditions and branches.
+  // Conditions and branches for MIPSr6.
+  void cmp(FPUCondition cond, SecondaryField fmt,
+         FPURegister fd, FPURegister ft, FPURegister fs);
+
+  void bc1eqz(int16_t offset, FPURegister ft);
+  void bc1eqz(Label* L, FPURegister ft) {
+    bc1eqz(branch_offset(L, false)>>2, ft);
+  }
+  void bc1nez(int16_t offset, FPURegister ft);
+  void bc1nez(Label* L, FPURegister ft) {
+    bc1nez(branch_offset(L, false)>>2, ft);
+  }
+
+  // Conditions and branches for non MIPSr6.
   void c(FPUCondition cond, SecondaryField fmt,
          FPURegister ft, FPURegister fs, uint16_t cc = 0);
 
   void bc1f(int16_t offset, uint16_t cc = 0);
-  void bc1f(Label* L, uint16_t cc = 0) { bc1f(branch_offset(L, false)>>2, cc); }
+  void bc1f(Label* L, uint16_t cc = 0) {
+    bc1f(branch_offset(L, false)>>2, cc);
+  }
   void bc1t(int16_t offset, uint16_t cc = 0);
-  void bc1t(Label* L, uint16_t cc = 0) { bc1t(branch_offset(L, false)>>2, cc); }
+  void bc1t(Label* L, uint16_t cc = 0) {
+    bc1t(branch_offset(L, false)>>2, cc);
+  }
   void fcmp(FPURegister src1, const double src2, FPUCondition cond);
 
   // Check the code size generated from label to here.
diff --git a/src/mips64/code-stubs-mips64.cc b/src/mips64/code-stubs-mips64.cc
index 6205e98cdd..f8b34cbd60 100644
--- a/src/mips64/code-stubs-mips64.cc
+++ b/src/mips64/code-stubs-mips64.cc
@@ -1016,17 +1016,28 @@ void ICCompareStub::GenerateGeneric(MacroAssembler* masm) {
 
   // Check if LESS condition is satisfied. If true, move conditionally
   // result to v0.
-  __ c(OLT, D, f12, f14);
-  __ Movt(v0, a4);
-  // Use previous check to store conditionally to v0 oposite condition
-  // (GREATER). If rhs is equal to lhs, this will be corrected in next
-  // check.
-  __ Movf(v0, a5);
-  // Check if EQUAL condition is satisfied. If true, move conditionally
-  // result to v0.
-  __ c(EQ, D, f12, f14);
-  __ Movt(v0, a6);
+  if (kArchVariant != kMips64r6) {
+    __ c(OLT, D, f12, f14);
+    __ Movt(v0, a4);
+    // Use previous check to store conditionally to v0 oposite condition
+    // (GREATER). If rhs is equal to lhs, this will be corrected in next
+    // check.
+    __ Movf(v0, a5);
+    // Check if EQUAL condition is satisfied. If true, move conditionally
+    // result to v0.
+    __ c(EQ, D, f12, f14);
+    __ Movt(v0, a6);
+  } else {
+    Label skip;
+    __ BranchF(USE_DELAY_SLOT, &skip, NULL, lt, f12, f14);
+    __ mov(v0, a4);  // Return LESS as result.
 
+    __ BranchF(USE_DELAY_SLOT, &skip, NULL, eq, f12, f14);
+    __ mov(v0, a6);  // Return EQUAL as result.
+
+    __ mov(v0, a5);  // Return GREATER as result.
+    __ bind(&skip);
+  }
   __ Ret();
 
   __ bind(&nan);
diff --git a/src/mips64/constants-mips64.cc b/src/mips64/constants-mips64.cc
index 04f4bbc52f..dfd62430c2 100644
--- a/src/mips64/constants-mips64.cc
+++ b/src/mips64/constants-mips64.cc
@@ -295,6 +295,8 @@ Instruction::Type Instruction::InstructionType() const {
     case COP1:    // Coprocessor instructions.
       switch (RsFieldRawNoAssert()) {
         case BC1:   // Branch on coprocessor condition.
+        case BC1EQZ:
+        case BC1NEZ:
           return kImmediateType;
         default:
           return kRegisterType;
@@ -322,6 +324,8 @@ Instruction::Type Instruction::InstructionType() const {
     case BNEL:
     case BLEZL:
     case BGTZL:
+    case BEQZC:
+    case BNEZC:
     case LB:
     case LH:
     case LWL:
diff --git a/src/mips64/constants-mips64.h b/src/mips64/constants-mips64.h
index 38e5aa3dd1..d3dd31da58 100644
--- a/src/mips64/constants-mips64.h
+++ b/src/mips64/constants-mips64.h
@@ -17,21 +17,17 @@
 #define UNSUPPORTED_MIPS() v8::internal::PrintF("Unsupported instruction.\n")
 
 enum ArchVariants {
-  kMips32r2,
-  kMips32r1,
-  kLoongson,
-  kMips64r2
+  kMips64r2,
+  kMips64r6
 };
 
 
 #ifdef _MIPS_ARCH_MIPS64R2
   static const ArchVariants kArchVariant = kMips64r2;
-#elif _MIPS_ARCH_LOONGSON
-// The loongson flag refers to the LOONGSON architectures based on MIPS-III,
-// which predates (and is a subset of) the mips32r2 and r1 architectures.
-  static const ArchVariants kArchVariant = kLoongson;
+#elif  _MIPS_ARCH_MIPS64R6
+  static const ArchVariants kArchVariant = kMips64r6;
 #else
-  static const ArchVariants kArchVariant = kMips64r1;
+  static const ArchVariants kArchVariant = kMips64r2;
 #endif
 
 
@@ -228,6 +224,8 @@ const int kLuiShift      = 16;
 
 const int kImm16Shift = 0;
 const int kImm16Bits  = 16;
+const int kImm21Shift = 0;
+const int kImm21Bits  = 21;
 const int kImm26Shift = 0;
 const int kImm26Bits  = 26;
 const int kImm28Shift = 0;
@@ -295,15 +293,17 @@ enum Opcode {
   ANDI      =   ((1 << 3) + 4) << kOpcodeShift,
   ORI       =   ((1 << 3) + 5) << kOpcodeShift,
   XORI      =   ((1 << 3) + 6) << kOpcodeShift,
-  LUI       =   ((1 << 3) + 7) << kOpcodeShift,
+  LUI       =   ((1 << 3) + 7) << kOpcodeShift,  // LUI/AUI family.
+  DAUI      =   ((3 << 3) + 5) << kOpcodeShift,
 
+  BEQC      =   ((2 << 3) + 0) << kOpcodeShift,
   COP1      =   ((2 << 3) + 1) << kOpcodeShift,  // Coprocessor 1 class.
   BEQL      =   ((2 << 3) + 4) << kOpcodeShift,
   BNEL      =   ((2 << 3) + 5) << kOpcodeShift,
   BLEZL     =   ((2 << 3) + 6) << kOpcodeShift,
   BGTZL     =   ((2 << 3) + 7) << kOpcodeShift,
 
-  DADDI     =   ((3 << 3) + 0) << kOpcodeShift,
+  DADDI     =   ((3 << 3) + 0) << kOpcodeShift,  // This is also BNEC.
   DADDIU    =   ((3 << 3) + 1) << kOpcodeShift,
   LDL       =   ((3 << 3) + 2) << kOpcodeShift,
   LDR       =   ((3 << 3) + 3) << kOpcodeShift,
@@ -330,6 +330,7 @@ enum Opcode {
   LWC1      =   ((6 << 3) + 1) << kOpcodeShift,
   LLD       =   ((6 << 3) + 4) << kOpcodeShift,
   LDC1      =   ((6 << 3) + 5) << kOpcodeShift,
+  BEQZC     =   ((6 << 3) + 6) << kOpcodeShift,
   LD        =   ((6 << 3) + 7) << kOpcodeShift,
 
   PREF      =   ((6 << 3) + 3) << kOpcodeShift,
@@ -337,6 +338,7 @@ enum Opcode {
   SWC1      =   ((7 << 3) + 1) << kOpcodeShift,
   SCD       =   ((7 << 3) + 4) << kOpcodeShift,
   SDC1      =   ((7 << 3) + 5) << kOpcodeShift,
+  BNEZC     =   ((7 << 3) + 6) << kOpcodeShift,
   SD        =   ((7 << 3) + 7) << kOpcodeShift,
 
   COP1X     =   ((1 << 4) + 3) << kOpcodeShift
@@ -359,6 +361,8 @@ enum SecondaryField {
   BREAK     =   ((1 << 3) + 5),
 
   MFHI      =   ((2 << 3) + 0),
+  CLZ_R6    =   ((2 << 3) + 0),
+  CLO_R6    =   ((2 << 3) + 1),
   MFLO      =   ((2 << 3) + 2),
   DSLLV     =   ((2 << 3) + 4),
   DSRLV     =   ((2 << 3) + 6),
@@ -394,7 +398,9 @@ enum SecondaryField {
   TLT       =   ((6 << 3) + 2),
   TLTU      =   ((6 << 3) + 3),
   TEQ       =   ((6 << 3) + 4),
+  SELEQZ_S  =   ((6 << 3) + 5),
   TNE       =   ((6 << 3) + 6),
+  SELNEZ_S  =   ((6 << 3) + 7),
 
   DSLL      =   ((7 << 3) + 0),
   DSRL      =   ((7 << 3) + 2),
@@ -402,6 +408,23 @@ enum SecondaryField {
   DSLL32    =   ((7 << 3) + 4),
   DSRL32    =   ((7 << 3) + 6),
   DSRA32    =   ((7 << 3) + 7),
+
+  // Multiply integers in r6.
+  MUL_MUH   =   ((3 << 3) + 0),  // MUL, MUH.
+  MUL_MUH_U =   ((3 << 3) + 1),  // MUL_U, MUH_U.
+  D_MUL_MUH =   ((7 << 2) + 0),  // DMUL, DMUH.
+  D_MUL_MUH_U = ((7 << 2) + 1),  // DMUL_U, DMUH_U.
+
+  MUL_OP    =   ((0 << 3) + 2),
+  MUH_OP    =   ((0 << 3) + 3),
+  DIV_OP    =   ((0 << 3) + 2),
+  MOD_OP    =   ((0 << 3) + 3),
+
+  DIV_MOD   =   ((3 << 3) + 2),
+  DIV_MOD_U =   ((3 << 3) + 3),
+  D_DIV_MOD =   ((3 << 3) + 6),
+  D_DIV_MOD_U = ((3 << 3) + 7),
+
   // drotr in special4?
 
   // SPECIAL2 Encoding of Function Field.
@@ -426,6 +449,9 @@ enum SecondaryField {
   BGEZ      =   ((0 << 3) + 1) << 16,
   BLTZAL    =   ((2 << 3) + 0) << 16,
   BGEZAL    =   ((2 << 3) + 1) << 16,
+  BGEZALL   =   ((2 << 3) + 3) << 16,
+  DAHI      =   ((0 << 3) + 6) << 16,
+  DATI      =   ((3 << 3) + 6) << 16,
 
   // COP1 Encoding of rs Field.
   MFC1      =   ((0 << 3) + 0) << 21,
@@ -472,6 +498,10 @@ enum SecondaryField {
   TRUNC_W_D =   ((1 << 3) + 5),
   CEIL_W_D  =   ((1 << 3) + 6),
   FLOOR_W_D =   ((1 << 3) + 7),
+  MIN       =   ((3 << 3) + 4),
+  MINA      =   ((3 << 3) + 5),
+  MAX       =   ((3 << 3) + 6),
+  MAXA      =   ((3 << 3) + 7),
   CVT_S_D   =   ((4 << 3) + 0),
   CVT_W_D   =   ((4 << 3) + 4),
   CVT_L_D   =   ((4 << 3) + 5),
@@ -488,6 +518,47 @@ enum SecondaryField {
   CVT_D_W   =   ((4 << 3) + 1),
   CVT_S_L   =   ((4 << 3) + 0),
   CVT_D_L   =   ((4 << 3) + 1),
+  BC1EQZ    =   ((2 << 2) + 1) << 21,
+  BC1NEZ    =   ((3 << 2) + 1) << 21,
+  // COP1 CMP positive predicates Bit 5..4 = 00.
+  CMP_AF    =   ((0 << 3) + 0),
+  CMP_UN    =   ((0 << 3) + 1),
+  CMP_EQ    =   ((0 << 3) + 2),
+  CMP_UEQ   =   ((0 << 3) + 3),
+  CMP_LT    =   ((0 << 3) + 4),
+  CMP_ULT   =   ((0 << 3) + 5),
+  CMP_LE    =   ((0 << 3) + 6),
+  CMP_ULE   =   ((0 << 3) + 7),
+  CMP_SAF   =   ((1 << 3) + 0),
+  CMP_SUN   =   ((1 << 3) + 1),
+  CMP_SEQ   =   ((1 << 3) + 2),
+  CMP_SUEQ  =   ((1 << 3) + 3),
+  CMP_SSLT  =   ((1 << 3) + 4),
+  CMP_SSULT =   ((1 << 3) + 5),
+  CMP_SLE   =   ((1 << 3) + 6),
+  CMP_SULE  =   ((1 << 3) + 7),
+  // COP1 CMP negative predicates Bit 5..4 = 01.
+  CMP_AT    =   ((2 << 3) + 0),  // Reserved, not implemented.
+  CMP_OR    =   ((2 << 3) + 1),
+  CMP_UNE   =   ((2 << 3) + 2),
+  CMP_NE    =   ((2 << 3) + 3),
+  CMP_UGE   =   ((2 << 3) + 4),  // Reserved, not implemented.
+  CMP_OGE   =   ((2 << 3) + 5),  // Reserved, not implemented.
+  CMP_UGT   =   ((2 << 3) + 6),  // Reserved, not implemented.
+  CMP_OGT   =   ((2 << 3) + 7),  // Reserved, not implemented.
+  CMP_SAT   =   ((3 << 3) + 0),  // Reserved, not implemented.
+  CMP_SOR   =   ((3 << 3) + 1),
+  CMP_SUNE  =   ((3 << 3) + 2),
+  CMP_SNE   =   ((3 << 3) + 3),
+  CMP_SUGE  =   ((3 << 3) + 4),  // Reserved, not implemented.
+  CMP_SOGE  =   ((3 << 3) + 5),  // Reserved, not implemented.
+  CMP_SUGT  =   ((3 << 3) + 6),  // Reserved, not implemented.
+  CMP_SOGT  =   ((3 << 3) + 7),  // Reserved, not implemented.
+
+  SEL       =   ((2 << 3) + 0),
+  SELEQZ_C  =   ((2 << 3) + 4),  // COP1 on FPR registers.
+  SELNEZ_C  =   ((2 << 3) + 7),  // COP1 on FPR registers.
+
   // COP1 Encoding of Function Field When rs=PS.
   // COP1X Encoding of Function Field.
   MADD_D    =   ((4 << 3) + 1),
@@ -497,9 +568,9 @@ enum SecondaryField {
 
 
 // ----- Emulated conditions.
-// On MIPS we use this enum to abstract from conditionnal branch instructions.
+// On MIPS we use this enum to abstract from conditional branch instructions.
 // The 'U' prefix is used to specify unsigned comparisons.
-// Oppposite conditions must be paired as odd/even numbers
+// Opposite conditions must be paired as odd/even numbers
 // because 'NegateCondition' function flips LSB to negate condition.
 enum Condition {
   // Any value < 0 is considered no_condition.
@@ -833,6 +904,11 @@ class Instruction {
     return Bits(kImm16Shift + kImm16Bits - 1, kImm16Shift);
   }
 
+  inline int32_t Imm21Value() const {
+    ASSERT(InstructionType() == kImmediateType);
+    return Bits(kImm21Shift + kImm21Bits - 1, kImm21Shift);
+  }
+
   inline int32_t Imm26Value() const {
     ASSERT(InstructionType() == kJumpType);
     return Bits(kImm26Shift + kImm26Bits - 1, kImm26Shift);
diff --git a/src/mips64/disasm-mips64.cc b/src/mips64/disasm-mips64.cc
index 9b9cd348f7..f2aaa58bef 100644
--- a/src/mips64/disasm-mips64.cc
+++ b/src/mips64/disasm-mips64.cc
@@ -86,6 +86,7 @@ class Decoder {
   void PrintUImm16(Instruction* instr);
   void PrintSImm16(Instruction* instr);
   void PrintXImm16(Instruction* instr);
+  void PrintXImm21(Instruction* instr);
   void PrintXImm26(Instruction* instr);
   void PrintCode(Instruction* instr);   // For break and trap instructions.
   // Printing of instruction name.
@@ -247,6 +248,13 @@ void Decoder::PrintXImm16(Instruction* instr) {
 }
 
 
+// Print 21-bit immediate value.
+void Decoder::PrintXImm21(Instruction* instr) {
+  uint32_t imm = instr->Imm21Value();
+  out_buffer_pos_ += SNPrintF(out_buffer_ + out_buffer_pos_, "0x%x", imm);
+}
+
+
 // Print 26-bit immediate value.
 void Decoder::PrintXImm26(Instruction* instr) {
   uint32_t imm = instr->Imm26Value() << kImmFieldShift;
@@ -361,7 +369,11 @@ int Decoder::FormatOption(Instruction* instr, const char* format) {
           PrintXImm16(instr);
         }
         return 6;
-      } else {
+      } else if (format[3] == '2' && format[4] == '1') {
+        ASSERT(STRING_STARTS_WITH(format, "imm21x"));
+        PrintXImm21(instr);
+        return 6;
+      } else if (format[3] == '2' && format[4] == '6') {
         ASSERT(STRING_STARTS_WITH(format, "imm26x"));
         PrintXImm26(instr);
         return 6;
@@ -466,9 +478,6 @@ int Decoder::DecodeTypeRegister(Instruction* instr) {
   switch (instr->OpcodeFieldRaw()) {
     case COP1:    // Coprocessor instructions.
       switch (instr->RsFieldRaw()) {
-        case BC1:   // bc1 handled in DecodeTypeImmediate.
-          UNREACHABLE();
-          break;
         case MFC1:
           Format(instr, "mfc1    'rt, 'fs");
           break;
@@ -582,14 +591,8 @@ int Decoder::DecodeTypeRegister(Instruction* instr) {
               break;
           }
           break;
-        case S:
-          UNIMPLEMENTED_MIPS();
-          break;
         case W:
           switch (instr->FunctionFieldRaw()) {
-            case CVT_S_W:   // Convert word to float (single).
-              Format(instr, "cvt.s.w 'fd, 'fs");
-              break;
             case CVT_D_W:   // Convert word to double.
               Format(instr, "cvt.d.w 'fd, 'fs");
               break;
@@ -605,13 +608,40 @@ int Decoder::DecodeTypeRegister(Instruction* instr) {
             case CVT_S_L:
               Format(instr, "cvt.s.l 'fd, 'fs");
               break;
+            case CMP_UN:
+              Format(instr, "cmp.un.d  'fd,  'fs, 'ft");
+              break;
+            case CMP_EQ:
+              Format(instr, "cmp.eq.d  'fd,  'fs, 'ft");
+              break;
+            case CMP_UEQ:
+              Format(instr, "cmp.ueq.d  'fd,  'fs, 'ft");
+              break;
+            case CMP_LT:
+              Format(instr, "cmp.lt.d  'fd,  'fs, 'ft");
+              break;
+            case CMP_ULT:
+              Format(instr, "cmp.ult.d  'fd,  'fs, 'ft");
+              break;
+            case CMP_LE:
+              Format(instr, "cmp.le.d  'fd,  'fs, 'ft");
+              break;
+            case CMP_ULE:
+              Format(instr, "cmp.ule.d  'fd,  'fs, 'ft");
+              break;
+            case CMP_OR:
+              Format(instr, "cmp.or.d  'fd,  'fs, 'ft");
+              break;
+            case CMP_UNE:
+              Format(instr, "cmp.une.d  'fd,  'fs, 'ft");
+              break;
+            case CMP_NE:
+              Format(instr, "cmp.ne.d  'fd,  'fs, 'ft");
+              break;
             default:
               UNREACHABLE();
           }
           break;
-        case PS:
-          UNIMPLEMENTED_MIPS();
-          break;
         default:
           UNREACHABLE();
       }
@@ -634,13 +664,24 @@ int Decoder::DecodeTypeRegister(Instruction* instr) {
           Format(instr, "jalr    'rs");
           break;
         case SLL:
-          if ( 0x0 == static_cast<int>(instr->InstructionBits()))
+          if (0x0 == static_cast<int>(instr->InstructionBits()))
             Format(instr, "nop");
           else
             Format(instr, "sll     'rd, 'rt, 'sa");
           break;
         case DSLL:
-          Format(instr, "dsll    'rd, 'rt, 'sa");
+            Format(instr, "dsll    'rd, 'rt, 'sa");
+          break;
+        case D_MUL_MUH:  // Equals to DMUL.
+          if (kArchVariant != kMips64r6) {
+            Format(instr, "dmult   'rs, 'rt");
+          } else {
+            if (instr->SaValue() == MUL_OP) {
+              Format(instr, "dmul   'rd, 'rs, 'rt");
+            } else {
+              Format(instr, "dmuh   'rd, 'rs, 'rt");
+            }
+          }
           break;
         case DSLL32:
           Format(instr, "dsll32  'rd, 'rt, 'sa");
@@ -714,34 +755,98 @@ int Decoder::DecodeTypeRegister(Instruction* instr) {
           Format(instr, "dsrav   'rd, 'rt, 'rs");
           break;
         case MFHI:
-          Format(instr, "mfhi    'rd");
+          if (instr->Bits(25, 16) == 0) {
+            Format(instr, "mfhi    'rd");
+          } else {
+            if ((instr->FunctionFieldRaw() == CLZ_R6)
+                && (instr->FdValue() == 1)) {
+              Format(instr, "clz     'rd, 'rs");
+            } else if ((instr->FunctionFieldRaw() == CLO_R6)
+                && (instr->FdValue() == 1)) {
+              Format(instr, "clo     'rd, 'rs");
+            }
+          }
           break;
         case MFLO:
           Format(instr, "mflo    'rd");
           break;
-        case MULT:
-          Format(instr, "mult    'rs, 'rt");
+        case D_MUL_MUH_U:  // Equals to DMULTU.
+          if (kArchVariant != kMips64r6) {
+              Format(instr, "dmultu  'rs, 'rt");
+          } else {
+            if (instr->SaValue() == MUL_OP) {
+              Format(instr, "dmulu  'rd, 'rs, 'rt");
+            } else {
+              Format(instr, "dmuhu  'rd, 'rs, 'rt");
+            }
+          }
           break;
-        case DMULT:
-          Format(instr, "dmult   'rs, 'rt");
+        case MULT:  // @Mips64r6 == MUL_MUH.
+          if (kArchVariant != kMips64r6) {
+            Format(instr, "mult    'rs, 'rt");
+          } else {
+            if (instr->SaValue() == MUL_OP) {
+              Format(instr, "mul    'rd, 'rs, 'rt");
+            } else {
+              Format(instr, "muh    'rd, 'rs, 'rt");
+            }
+          }
           break;
-        case MULTU:
-          Format(instr, "multu   'rs, 'rt");
+        case MULTU:  // @Mips64r6 == MUL_MUH_U.
+          if (kArchVariant != kMips64r6) {
+            Format(instr, "multu   'rs, 'rt");
+          } else {
+            if (instr->SaValue() == MUL_OP) {
+              Format(instr, "mulu   'rd, 'rs, 'rt");
+            } else {
+              Format(instr, "muhu   'rd, 'rs, 'rt");
+            }
+          }
+
           break;
-        case DMULTU:
-          Format(instr, "dmultu  'rs, 'rt");
+        case DIV:  // @Mips64r6 == DIV_MOD.
+          if (kArchVariant != kMips64r6) {
+            Format(instr, "div     'rs, 'rt");
+          } else {
+            if (instr->SaValue() == DIV_OP) {
+              Format(instr, "div    'rd, 'rs, 'rt");
+            } else {
+              Format(instr, "mod    'rd, 'rs, 'rt");
+            }
+          }
           break;
-        case DIV:
-          Format(instr, "div     'rs, 'rt");
+        case DDIV:  // @Mips64r6 == D_DIV_MOD.
+          if (kArchVariant != kMips64r6) {
+            Format(instr, "ddiv    'rs, 'rt");
+          } else {
+            if (instr->SaValue() == DIV_OP) {
+              Format(instr, "ddiv   'rd, 'rs, 'rt");
+            } else {
+              Format(instr, "dmod   'rd, 'rs, 'rt");
+            }
+          }
           break;
-        case DDIV:
-          Format(instr, "ddiv    'rs, 'rt");
+        case DIVU:  // @Mips64r6 == DIV_MOD_U.
+          if (kArchVariant != kMips64r6) {
+            Format(instr, "divu    'rs, 'rt");
+          } else {
+            if (instr->SaValue() == DIV_OP) {
+              Format(instr, "divu   'rd, 'rs, 'rt");
+            } else {
+              Format(instr, "modu   'rd, 'rs, 'rt");
+            }
+          }
           break;
-        case DIVU:
-          Format(instr, "divu    'rs, 'rt");
-          break;
-        case DDIVU:
-          Format(instr, "ddivu   'rs, 'rt");
+        case DDIVU:  // @Mips64r6 == D_DIV_MOD_U.
+          if (kArchVariant != kMips64r6) {
+            Format(instr, "ddivu   'rs, 'rt");
+          } else {
+            if (instr->SaValue() == DIV_OP) {
+              Format(instr, "ddivu  'rd, 'rs, 'rt");
+            } else {
+              Format(instr, "dmodu  'rd, 'rs, 'rt");
+            }
+          }
           break;
         case ADD:
           Format(instr, "add     'rd, 'rs, 'rt");
@@ -824,6 +929,12 @@ int Decoder::DecodeTypeRegister(Instruction* instr) {
             Format(instr, "movf    'rd, 'rs, 'bc");
           }
           break;
+        case SELEQZ_S:
+          Format(instr, "seleqz    'rd, 'rs, 'rt");
+          break;
+        case SELNEZ_S:
+          Format(instr, "selnez    'rd, 'rs, 'rt");
+          break;
         default:
           UNREACHABLE();
       }
@@ -834,7 +945,9 @@ int Decoder::DecodeTypeRegister(Instruction* instr) {
           Format(instr, "mul     'rd, 'rs, 'rt");
           break;
         case CLZ:
-          Format(instr, "clz     'rd, 'rs");
+          if (kArchVariant != kMips64r6) {
+            Format(instr, "clz     'rd, 'rs");
+          }
           break;
         default:
           UNREACHABLE();
@@ -843,19 +956,11 @@ int Decoder::DecodeTypeRegister(Instruction* instr) {
     case SPECIAL3:
       switch (instr->FunctionFieldRaw()) {
         case INS: {
-          if (kArchVariant == kMips64r2) {
-            Format(instr, "ins     'rt, 'rs, 'sa, 'ss2");
-          } else {
-            Unknown(instr);
-          }
+          Format(instr, "ins     'rt, 'rs, 'sa, 'ss2");
           break;
         }
         case EXT: {
-          if (kArchVariant == kMips64r2) {
-            Format(instr, "ext     'rt, 'rs, 'sa, 'ss1");
-          } else {
-            Unknown(instr);
-          }
+          Format(instr, "ext     'rt, 'rs, 'sa, 'ss1");
           break;
         }
         default:
@@ -871,7 +976,6 @@ int Decoder::DecodeTypeRegister(Instruction* instr) {
 
 void Decoder::DecodeTypeImmediate(Instruction* instr) {
   switch (instr->OpcodeFieldRaw()) {
-    // ------------- REGIMM class.
     case COP1:
       switch (instr->RsFieldRaw()) {
         case BC1:
@@ -881,10 +985,150 @@ void Decoder::DecodeTypeImmediate(Instruction* instr) {
             Format(instr, "bc1f    'bc, 'imm16u");
           }
           break;
+        case BC1EQZ:
+          Format(instr, "bc1eqz    'ft, 'imm16u");
+          break;
+        case BC1NEZ:
+          Format(instr, "bc1nez    'ft, 'imm16u");
+          break;
+        case W:  // CMP.S instruction.
+          switch (instr->FunctionValue()) {
+            case CMP_AF:
+              Format(instr, "cmp.af.S    'ft, 'fs, 'fd");
+              break;
+            case CMP_UN:
+              Format(instr, "cmp.un.S    'ft, 'fs, 'fd");
+              break;
+            case CMP_EQ:
+              Format(instr, "cmp.eq.S    'ft, 'fs, 'fd");
+              break;
+            case CMP_UEQ:
+              Format(instr, "cmp.ueq.S   'ft, 'fs, 'fd");
+              break;
+            case CMP_LT:
+              Format(instr, "cmp.lt.S    'ft, 'fs, 'fd");
+              break;
+            case CMP_ULT:
+              Format(instr, "cmp.ult.S   'ft, 'fs, 'fd");
+              break;
+            case CMP_LE:
+              Format(instr, "cmp.le.S    'ft, 'fs, 'fd");
+              break;
+            case CMP_ULE:
+              Format(instr, "cmp.ule.S   'ft, 'fs, 'fd");
+              break;
+            case CMP_OR:
+              Format(instr, "cmp.or.S    'ft, 'fs, 'fd");
+              break;
+            case CMP_UNE:
+              Format(instr, "cmp.une.S   'ft, 'fs, 'fd");
+              break;
+            case CMP_NE:
+              Format(instr, "cmp.ne.S    'ft, 'fs, 'fd");
+              break;
+            default:
+              UNREACHABLE();
+          }
+          break;
+        case L:  // CMP.D instruction.
+          switch (instr->FunctionValue()) {
+            case CMP_AF:
+              Format(instr, "cmp.af.D    'ft, 'fs, 'fd");
+              break;
+            case CMP_UN:
+              Format(instr, "cmp.un.D    'ft, 'fs, 'fd");
+              break;
+            case CMP_EQ:
+              Format(instr, "cmp.eq.D    'ft, 'fs, 'fd");
+              break;
+            case CMP_UEQ:
+              Format(instr, "cmp.ueq.D   'ft, 'fs, 'fd");
+              break;
+            case CMP_LT:
+              Format(instr, "cmp.lt.D    'ft, 'fs, 'fd");
+              break;
+            case CMP_ULT:
+              Format(instr, "cmp.ult.D   'ft, 'fs, 'fd");
+              break;
+            case CMP_LE:
+              Format(instr, "cmp.le.D    'ft, 'fs, 'fd");
+              break;
+            case CMP_ULE:
+              Format(instr, "cmp.ule.D   'ft, 'fs, 'fd");
+              break;
+            case CMP_OR:
+              Format(instr, "cmp.or.D    'ft, 'fs, 'fd");
+              break;
+            case CMP_UNE:
+              Format(instr, "cmp.une.D   'ft, 'fs, 'fd");
+              break;
+            case CMP_NE:
+              Format(instr, "cmp.ne.D    'ft, 'fs, 'fd");
+              break;
+            default:
+              UNREACHABLE();
+          }
+          break;
+        case S:
+          switch (instr->FunctionValue()) {
+            case SEL:
+              Format(instr, "sel.S    'ft, 'fs, 'fd");
+              break;
+            case SELEQZ_C:
+              Format(instr, "seleqz.S 'ft, 'fs, 'fd");
+              break;
+            case SELNEZ_C:
+              Format(instr, "selnez.S 'ft, 'fs, 'fd");
+              break;
+            case MIN:
+              Format(instr, "min.S    'ft, 'fs, 'fd");
+              break;
+            case MINA:
+              Format(instr, "mina.S   'ft, 'fs, 'fd");
+              break;
+            case MAX:
+              Format(instr, "max.S    'ft, 'fs, 'fd");
+              break;
+            case MAXA:
+              Format(instr, "maxa.S   'ft, 'fs, 'fd");
+              break;
+            default:
+              UNREACHABLE();
+          }
+          break;
+        case D:
+          switch (instr->FunctionValue()) {
+            case SEL:
+              Format(instr, "sel.D    'ft, 'fs, 'fd");
+              break;
+            case SELEQZ_C:
+              Format(instr, "seleqz.D 'ft, 'fs, 'fd");
+              break;
+            case SELNEZ_C:
+              Format(instr, "selnez.D 'ft, 'fs, 'fd");
+              break;
+            case MIN:
+              Format(instr, "min.D    'ft, 'fs, 'fd");
+              break;
+            case MINA:
+              Format(instr, "mina.D   'ft, 'fs, 'fd");
+              break;
+            case MAX:
+              Format(instr, "max.D    'ft, 'fs, 'fd");
+              break;
+            case MAXA:
+              Format(instr, "maxa.D   'ft, 'fs, 'fd");
+              break;
+            default:
+              UNREACHABLE();
+          }
+          break;
         default:
           UNREACHABLE();
       }
+
       break;  // Case COP1.
+    // ------------- REGIMM class.
     case REGIMM:
       switch (instr->RtFieldRaw()) {
         case BLTZ:
@@ -899,6 +1143,15 @@ void Decoder::DecodeTypeImmediate(Instruction* instr) {
         case BGEZAL:
           Format(instr, "bgezal  'rs, 'imm16u");
           break;
+        case BGEZALL:
+          Format(instr, "bgezall 'rs, 'imm16u");
+          break;
+        case DAHI:
+          Format(instr, "dahi    'rs, 'imm16u");
+          break;
+        case DATI:
+          Format(instr, "dati    'rs, 'imm16u");
+          break;
         default:
           UNREACHABLE();
       }
@@ -911,17 +1164,105 @@ void Decoder::DecodeTypeImmediate(Instruction* instr) {
       Format(instr, "bne     'rs, 'rt, 'imm16u");
       break;
     case BLEZ:
-      Format(instr, "blez    'rs, 'imm16u");
+      if ((instr->RtFieldRaw() == 0)
+          && (instr->RsFieldRaw() != 0)) {
+        Format(instr, "blez    'rs, 'imm16u");
+      } else if ((instr->RtFieldRaw() != instr->RsFieldRaw())
+          && (instr->RsFieldRaw() != 0) && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "bgeuc    'rs, 'rt, 'imm16u");
+      } else if ((instr->RtFieldRaw() == instr->RsFieldRaw())
+          && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "bgezalc  'rs, 'imm16u");
+      } else if ((instr->RsFieldRaw() == 0)
+          && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "blezalc  'rs, 'imm16u");
+      } else {
+        UNREACHABLE();
+      }
       break;
     case BGTZ:
-      Format(instr, "bgtz    'rs, 'imm16u");
+      if ((instr->RtFieldRaw() == 0)
+          && (instr->RsFieldRaw() != 0)) {
+        Format(instr, "bgtz    'rs, 'imm16u");
+      } else if ((instr->RtFieldRaw() != instr->RsFieldRaw())
+          && (instr->RsFieldRaw() != 0) && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "bltuc   'rs, 'rt, 'imm16u");
+      } else if ((instr->RtFieldRaw() == instr->RsFieldRaw())
+          && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "bltzalc 'rt, 'imm16u");
+      } else if ((instr->RsFieldRaw() == 0)
+          && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "bgtzalc 'rt, 'imm16u");
+      } else {
+        UNREACHABLE();
+      }
+      break;
+    case BLEZL:
+      if ((instr->RtFieldRaw() == instr->RsFieldRaw())
+          && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "bgezc    'rt, 'imm16u");
+      } else if ((instr->RtFieldRaw() != instr->RsFieldRaw())
+          && (instr->RsFieldRaw() != 0) && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "bgec     'rs, 'rt, 'imm16u");
+      } else if ((instr->RsFieldRaw() == 0)
+          && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "blezc    'rt, 'imm16u");
+      } else {
+        UNREACHABLE();
+      }
+      break;
+    case BGTZL:
+      if ((instr->RtFieldRaw() == instr->RsFieldRaw())
+          && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "bltzc    'rt, 'imm16u");
+      } else if ((instr->RtFieldRaw() != instr->RsFieldRaw())
+          && (instr->RsFieldRaw() != 0) && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "bltc     'rs, 'rt, 'imm16u");
+      } else if ((instr->RsFieldRaw() == 0)
+          && (instr->RtFieldRaw() != 0)) {
+        Format(instr, "bgtzc    'rt, 'imm16u");
+      } else {
+        UNREACHABLE();
+      }
+      break;
+    case BEQZC:
+      if (instr->RsFieldRaw() != 0) {
+        Format(instr, "beqzc   'rs, 'imm21x");
+      }
+      break;
+    case BNEZC:
+      if (instr->RsFieldRaw() != 0) {
+        Format(instr, "bnezc   'rs, 'imm21x");
+      }
       break;
     // ------------- Arithmetic instructions.
     case ADDI:
-      Format(instr, "addi    'rt, 'rs, 'imm16s");
+      if (kArchVariant != kMips64r6) {
+        Format(instr, "addi    'rt, 'rs, 'imm16s");
+      } else {
+        // Check if BOVC or BEQC instruction.
+        if (instr->RsFieldRaw() >= instr->RtFieldRaw()) {
+          Format(instr, "bovc  'rs, 'rt, 'imm16s");
+        } else if (instr->RsFieldRaw() < instr->RtFieldRaw()) {
+          Format(instr, "beqc  'rs, 'rt, 'imm16s");
+        } else {
+          UNREACHABLE();
+        }
+      }
       break;
     case DADDI:
-      Format(instr, "daddi   'rt, 'rs, 'imm16s");
+      if (kArchVariant != kMips64r6) {
+        Format(instr, "daddi   'rt, 'rs, 'imm16s");
+      } else {
+        // Check if BNVC or BNEC instruction.
+        if (instr->RsFieldRaw() >= instr->RtFieldRaw()) {
+          Format(instr, "bnvc  'rs, 'rt, 'imm16s");
+        } else if (instr->RsFieldRaw() < instr->RtFieldRaw()) {
+          Format(instr, "bnec  'rs, 'rt, 'imm16s");
+        } else {
+          UNREACHABLE();
+        }
+      }
       break;
     case ADDIU:
       Format(instr, "addiu   'rt, 'rs, 'imm16s");
@@ -945,7 +1286,18 @@ void Decoder::DecodeTypeImmediate(Instruction* instr) {
       Format(instr, "xori    'rt, 'rs, 'imm16x");
       break;
     case LUI:
-      Format(instr, "lui     'rt, 'imm16x");
+      if (kArchVariant != kMips64r6) {
+        Format(instr, "lui     'rt, 'imm16x");
+      } else {
+        if (instr->RsValue() != 0) {
+          Format(instr, "aui     'rt, 'imm16x");
+        } else {
+          Format(instr, "lui     'rt, 'imm16x");
+        }
+      }
+      break;
+    case DAUI:
+      Format(instr, "daui    'rt, 'imm16x");
       break;
     // ------------- Memory instructions.
     case LB:
diff --git a/src/mips64/full-codegen-mips64.cc b/src/mips64/full-codegen-mips64.cc
index 212d917e39..efdcdfb33c 100644
--- a/src/mips64/full-codegen-mips64.cc
+++ b/src/mips64/full-codegen-mips64.cc
@@ -2364,14 +2364,12 @@ void FullCodeGenerator::EmitInlineSmiBinaryOp(BinaryOperation* expr,
       __ BranchOnOverflow(&stub_call, scratch1);
       break;
     case Token::MUL: {
-      __ SmiUntag(scratch1, right);
-      __ Dmult(left, scratch1);
-      __ mflo(scratch1);
-      __ mfhi(scratch2);
-      __ dsra32(scratch1, scratch1, 31);
-      __ Branch(&stub_call, ne, scratch1, Operand(scratch2));
-      __ mflo(v0);
-      __ Branch(&done, ne, v0, Operand(zero_reg));
+      __ Dmulh(v0, left, right);
+      __ dsra32(scratch2, v0, 0);
+      __ sra(scratch1, v0, 31);
+      __ Branch(USE_DELAY_SLOT, &stub_call, ne, scratch2, Operand(scratch1));
+      __ SmiTag(v0);
+      __ Branch(USE_DELAY_SLOT, &done, ne, v0, Operand(zero_reg));
       __ Daddu(scratch2, right, left);
       __ Branch(&stub_call, lt, scratch2, Operand(zero_reg));
       ASSERT(Smi::FromInt(0) == 0);
@@ -3943,12 +3941,11 @@ void FullCodeGenerator::EmitFastAsciiArrayJoin(CallRuntime* expr) {
   __ ld(scratch1, FieldMemOperand(separator, SeqOneByteString::kLengthOffset));
   __ Dsubu(string_length, string_length, Operand(scratch1));
   __ SmiUntag(scratch1);
-  __ Dmult(array_length, scratch1);
+  __ Dmul(scratch2, array_length, scratch1);
   // Check for smi overflow. No overflow if higher 33 bits of 64-bit result are
   // zero.
-  __ mfhi(scratch2);
+  __ dsra32(scratch1, scratch2, 0);
   __ Branch(&bailout, ne, scratch2, Operand(zero_reg));
-  __ mflo(scratch2);
   __ SmiUntag(string_length);
   __ AdduAndCheckForOverflow(string_length, string_length, scratch2, scratch3);
   __ BranchOnOverflow(&bailout, scratch3);
diff --git a/src/mips64/lithium-codegen-mips64.cc b/src/mips64/lithium-codegen-mips64.cc
index 4a625610d8..d1bdacd680 100644
--- a/src/mips64/lithium-codegen-mips64.cc
+++ b/src/mips64/lithium-codegen-mips64.cc
@@ -1119,7 +1119,7 @@ void LCodeGen::DoModI(LModI* instr) {
   const Register result_reg = ToRegister(instr->result());
 
   // div runs in the background while we check for special cases.
-  __ ddiv(left_reg, right_reg);
+  __ Dmod(result_reg, left_reg, right_reg);
 
   Label done;
   // Check for x % 0, we have to deopt in this case because we can't return a
@@ -1144,8 +1144,7 @@ void LCodeGen::DoModI(LModI* instr) {
   }
 
   // If we care about -0, test if the dividend is <0 and the result is 0.
-  __ Branch(USE_DELAY_SLOT, &done, ge, left_reg, Operand(zero_reg));
-  __ mfhi(result_reg);
+  __ Branch(&done, ge, left_reg, Operand(zero_reg));
 
   if (hmod->CheckFlag(HValue::kBailoutOnMinusZero)) {
     DeoptimizeIf(eq, instr->environment(), result_reg, Operand(zero_reg));
@@ -1235,7 +1234,7 @@ void LCodeGen::DoDivI(LDivI* instr) {
 
   // On MIPS div is asynchronous - it will run in the background while we
   // check for special cases.
-  __ ddiv(dividend, divisor);
+  __ Ddiv(result, dividend, divisor);
 
   // Check for x / 0.
   if (hdiv->CheckFlag(HValue::kCanBeDivByZero)) {
@@ -1260,11 +1259,14 @@ void LCodeGen::DoDivI(LDivI* instr) {
   }
 
   if (!hdiv->CheckFlag(HValue::kAllUsesTruncatingToInt32)) {
-    __ mfhi(result);
-    DeoptimizeIf(ne, instr->environment(), result, Operand(zero_reg));
-    __ mflo(result);
-  } else {
-    __ mflo(result);
+    // Calculate remainder.
+    Register remainder = ToRegister(instr->temp());
+    if (kArchVariant != kMips64r6) {
+      __ mfhi(remainder);
+    } else {
+      __ dmod(remainder, dividend, divisor);
+    }
+    DeoptimizeIf(ne, instr->environment(), remainder, Operand(zero_reg));
   }
 }
 
@@ -1391,7 +1393,7 @@ void LCodeGen::DoFlooringDivI(LFlooringDivI* instr) {
 
   // On MIPS div is asynchronous - it will run in the background while we
   // check for special cases.
-  __ ddiv(dividend, divisor);
+  __ Ddiv(result, dividend, divisor);
 
   // Check for x / 0.
   if (hdiv->CheckFlag(HValue::kCanBeDivByZero)) {
@@ -1418,8 +1420,11 @@ void LCodeGen::DoFlooringDivI(LFlooringDivI* instr) {
   // We performed a truncating division. Correct the result if necessary.
   Label done;
   Register remainder = scratch0();
-  __ mfhi(remainder);
-  __ mflo(result);
+  if (kArchVariant != kMips64r6) {
+    __ mfhi(remainder);
+  } else {
+    __ dmod(remainder, dividend, divisor);
+  }
   __ Branch(&done, eq, remainder, Operand(zero_reg), USE_DELAY_SLOT);
   __ Xor(remainder, remainder, Operand(divisor));
   __ Branch(&done, ge, remainder, Operand(zero_reg));
@@ -1507,21 +1512,16 @@ void LCodeGen::DoMulI(LMulI* instr) {
     if (overflow) {
       // hi:lo = left * right.
       if (instr->hydrogen()->representation().IsSmi()) {
-        __ SmiUntag(result, left);
-        __ dmult(result, right);
-        __ mfhi(scratch);
-        __ mflo(result);
+        __ Dmulh(result, left, right);
       } else {
-        __ dmult(left, right);
-        __ mfhi(scratch);
-        __ mflo(result);
+        __ Dmul(result, left, right);
+      }
+      __ dsra32(scratch, result, 0);
+      __ sra(at, result, 31);
+      if (instr->hydrogen()->representation().IsSmi()) {
+        __ SmiTag(result);
       }
-      __ dsra32(at, result, 31);
       DeoptimizeIf(ne, instr->environment(), scratch, Operand(at));
-      if (!instr->hydrogen()->representation().IsSmi()) {
-        DeoptimizeIf(gt, instr->environment(), result, Operand(kMaxInt));
-        DeoptimizeIf(lt, instr->environment(), result, Operand(kMinInt));
-      }
     } else {
       if (instr->hydrogen()->representation().IsSmi()) {
         __ SmiUntag(result, left);
diff --git a/src/mips64/lithium-mips64.cc b/src/mips64/lithium-mips64.cc
index 1f5f578477..326ac8a794 100644
--- a/src/mips64/lithium-mips64.cc
+++ b/src/mips64/lithium-mips64.cc
@@ -1307,8 +1307,10 @@ LInstruction* LChunkBuilder::DoDivI(HDiv* instr) {
   ASSERT(instr->right()->representation().Equals(instr->representation()));
   LOperand* dividend = UseRegister(instr->left());
   LOperand* divisor = UseRegister(instr->right());
+  LOperand* temp = instr->CheckFlag(HInstruction::kAllUsesTruncatingToInt32)
+    ? NULL : TempRegister();
   LInstruction* result =
-      DefineAsRegister(new(zone()) LDivI(dividend, divisor));
+      DefineAsRegister(new(zone()) LDivI(dividend, divisor, temp));
   if (instr->CheckFlag(HValue::kCanBeDivByZero) ||
       instr->CheckFlag(HValue::kBailoutOnMinusZero) ||
       (instr->CheckFlag(HValue::kCanOverflow) &&
diff --git a/src/mips64/lithium-mips64.h b/src/mips64/lithium-mips64.h
index f58f0da22d..37f8efa5e2 100644
--- a/src/mips64/lithium-mips64.h
+++ b/src/mips64/lithium-mips64.h
@@ -689,15 +689,17 @@ class LDivByConstI V8_FINAL : public LTemplateInstruction<1, 1, 0> {
 };
 
 
-class LDivI V8_FINAL : public LTemplateInstruction<1, 2, 0> {
+class LDivI V8_FINAL : public LTemplateInstruction<1, 2, 1> {
  public:
-  LDivI(LOperand* dividend, LOperand* divisor) {
+  LDivI(LOperand* dividend, LOperand* divisor,  LOperand* temp) {
     inputs_[0] = dividend;
     inputs_[1] = divisor;
+    temps_[0] = temp;
   }
 
   LOperand* dividend() { return inputs_[0]; }
   LOperand* divisor() { return inputs_[1]; }
+  LOperand* temp() { return temps_[0]; }
 
   DECLARE_CONCRETE_INSTRUCTION(DivI, "div-i")
   DECLARE_HYDROGEN_ACCESSOR(BinaryOperation)
diff --git a/src/mips64/macro-assembler-mips64.cc b/src/mips64/macro-assembler-mips64.cc
index dea3433086..bf5a6647ce 100644
--- a/src/mips64/macro-assembler-mips64.cc
+++ b/src/mips64/macro-assembler-mips64.cc
@@ -676,21 +676,33 @@ void MacroAssembler::Dsubu(Register rd, Register rs, const Operand& rt) {
 
 void MacroAssembler::Mul(Register rd, Register rs, const Operand& rt) {
   if (rt.is_reg()) {
-    if (kArchVariant == kLoongson) {
+    mul(rd, rs, rt.rm());
+  } else {
+    // li handles the relocation.
+    ASSERT(!rs.is(at));
+    li(at, rt);
+    mul(rd, rs, at);
+  }
+}
+
+
+void MacroAssembler::Mulh(Register rd, Register rs, const Operand& rt) {
+  if (rt.is_reg()) {
+    if (kArchVariant != kMips64r6) {
       mult(rs, rt.rm());
-      mflo(rd);
+      mfhi(rd);
     } else {
-      mul(rd, rs, rt.rm());
+      muh(rd, rs, rt.rm());
     }
   } else {
     // li handles the relocation.
     ASSERT(!rs.is(at));
     li(at, rt);
-    if (kArchVariant == kLoongson) {
+    if (kArchVariant != kMips64r6) {
       mult(rs, at);
-      mflo(rd);
+      mfhi(rd);
     } else {
-      mul(rd, rs, at);
+      muh(rd, rs, at);
     }
   }
 }
@@ -698,12 +710,9 @@ void MacroAssembler::Mul(Register rd, Register rs, const Operand& rt) {
 
 void MacroAssembler::Dmul(Register rd, Register rs, const Operand& rt) {
   if (rt.is_reg()) {
-    if (kArchVariant == kLoongson) {
-      dmult(rs, rt.rm());
-      mflo(rd);
+    if (kArchVariant == kMips64r6) {
+      dmul(rd, rs, rt.rm());
     } else {
-      // TODO(yuyin):
-      // dmul(rd, rs, rt.rm());
       dmult(rs, rt.rm());
       mflo(rd);
     }
@@ -711,12 +720,9 @@ void MacroAssembler::Dmul(Register rd, Register rs, const Operand& rt) {
     // li handles the relocation.
     ASSERT(!rs.is(at));
     li(at, rt);
-    if (kArchVariant == kLoongson) {
-      dmult(rs, at);
-      mflo(rd);
+    if (kArchVariant == kMips64r6) {
+      dmul(rd, rs, at);
     } else {
-      // TODO(yuyin):
-      // dmul(rd, rs, at);
       dmult(rs, at);
       mflo(rd);
     }
@@ -724,6 +730,28 @@ void MacroAssembler::Dmul(Register rd, Register rs, const Operand& rt) {
 }
 
 
+void MacroAssembler::Dmulh(Register rd, Register rs, const Operand& rt) {
+  if (rt.is_reg()) {
+    if (kArchVariant == kMips64r6) {
+      dmuh(rd, rs, rt.rm());
+    } else {
+      dmult(rs, rt.rm());
+      mfhi(rd);
+    }
+  } else {
+    // li handles the relocation.
+    ASSERT(!rs.is(at));
+    li(at, rt);
+    if (kArchVariant == kMips64r6) {
+      dmuh(rd, rs, at);
+    } else {
+      dmult(rs, at);
+      mfhi(rd);
+    }
+  }
+}
+
+
 void MacroAssembler::Mult(Register rs, const Operand& rt) {
   if (rt.is_reg()) {
     mult(rs, rt.rm());
@@ -796,6 +824,31 @@ void MacroAssembler::Ddiv(Register rs, const Operand& rt) {
 }
 
 
+void MacroAssembler::Ddiv(Register rd, Register rs, const Operand& rt) {
+  if (kArchVariant != kMips64r6) {
+    if (rt.is_reg()) {
+      ddiv(rs, rt.rm());
+      mflo(rd);
+    } else {
+      // li handles the relocation.
+      ASSERT(!rs.is(at));
+      li(at, rt);
+      ddiv(rs, at);
+      mflo(rd);
+    }
+  } else {
+    if (rt.is_reg()) {
+      ddiv(rd, rs, rt.rm());
+    } else {
+      // li handles the relocation.
+      ASSERT(!rs.is(at));
+      li(at, rt);
+      ddiv(rd, rs, at);
+    }
+  }
+}
+
+
 void MacroAssembler::Divu(Register rs, const Operand& rt) {
   if (rt.is_reg()) {
     divu(rs, rt.rm());
@@ -820,6 +873,31 @@ void MacroAssembler::Ddivu(Register rs, const Operand& rt) {
 }
 
 
+void MacroAssembler::Dmod(Register rd, Register rs, const Operand& rt) {
+  if (kArchVariant != kMips64r6) {
+    if (rt.is_reg()) {
+      ddiv(rs, rt.rm());
+      mfhi(rd);
+    } else {
+      // li handles the relocation.
+      ASSERT(!rs.is(at));
+      li(at, rt);
+      ddiv(rs, at);
+      mfhi(rd);
+    }
+  } else {
+    if (rt.is_reg()) {
+      dmod(rd, rs, rt.rm());
+    } else {
+      // li handles the relocation.
+      ASSERT(!rs.is(at));
+      li(at, rt);
+      dmod(rd, rs, at);
+    }
+  }
+}
+
+
 void MacroAssembler::And(Register rd, Register rs, const Operand& rt) {
   if (rt.is_reg()) {
     and_(rd, rs, rt.rm());
@@ -957,11 +1035,7 @@ void MacroAssembler::Dror(Register rd, Register rs, const Operand& rt) {
 
 
 void MacroAssembler::Pref(int32_t hint, const MemOperand& rs) {
-  if (kArchVariant == kLoongson) {
-    lw(zero_reg, rs);
-  } else {
     pref(hint, rs);
-  }
 }
 
 
@@ -1385,49 +1459,102 @@ void MacroAssembler::BranchF(Label* target,
   ASSERT(nan || target);
   // Check for unordered (NaN) cases.
   if (nan) {
-    c(UN, D, cmp1, cmp2);
-    bc1t(nan);
+    if (kArchVariant != kMips64r6) {
+      c(UN, D, cmp1, cmp2);
+      bc1t(nan);
+    } else {
+      // Use f31 for comparison result. It has to be unavailable to lithium
+      // register allocator.
+      ASSERT(!cmp1.is(f31) && !cmp2.is(f31));
+      cmp(UN, L, f31, cmp1, cmp2);
+      bc1nez(nan, f31);
+    }
   }
 
-  if (target) {
-    // Here NaN cases were either handled by this function or are assumed to
-    // have been handled by the caller.
-    // Unsigned conditions are treated as their signed counterpart.
-    switch (cc) {
-      case lt:
-        c(OLT, D, cmp1, cmp2);
-        bc1t(target);
-        break;
-      case gt:
-        c(ULE, D, cmp1, cmp2);
-        bc1f(target);
-        break;
-      case ge:
-        c(ULT, D, cmp1, cmp2);
-        bc1f(target);
-        break;
-      case le:
-        c(OLE, D, cmp1, cmp2);
-        bc1t(target);
-        break;
-      case eq:
-        c(EQ, D, cmp1, cmp2);
-        bc1t(target);
-        break;
-      case ueq:
-        c(UEQ, D, cmp1, cmp2);
-        bc1t(target);
-        break;
-      case ne:
-        c(EQ, D, cmp1, cmp2);
-        bc1f(target);
-        break;
-      case nue:
-        c(UEQ, D, cmp1, cmp2);
-        bc1f(target);
-        break;
-      default:
-        CHECK(0);
+  if (kArchVariant != kMips64r6) {
+    if (target) {
+      // Here NaN cases were either handled by this function or are assumed to
+      // have been handled by the caller.
+      switch (cc) {
+        case lt:
+          c(OLT, D, cmp1, cmp2);
+          bc1t(target);
+          break;
+        case gt:
+          c(ULE, D, cmp1, cmp2);
+          bc1f(target);
+          break;
+        case ge:
+          c(ULT, D, cmp1, cmp2);
+          bc1f(target);
+          break;
+        case le:
+          c(OLE, D, cmp1, cmp2);
+          bc1t(target);
+          break;
+        case eq:
+          c(EQ, D, cmp1, cmp2);
+          bc1t(target);
+          break;
+        case ueq:
+          c(UEQ, D, cmp1, cmp2);
+          bc1t(target);
+          break;
+        case ne:
+          c(EQ, D, cmp1, cmp2);
+          bc1f(target);
+          break;
+        case nue:
+          c(UEQ, D, cmp1, cmp2);
+          bc1f(target);
+          break;
+        default:
+          CHECK(0);
+      }
+    }
+  } else {
+    if (target) {
+      // Here NaN cases were either handled by this function or are assumed to
+      // have been handled by the caller.
+      // Unsigned conditions are treated as their signed counterpart.
+      // Use f31 for comparison result, it is valid in fp64 (FR = 1) mode.
+      ASSERT(!cmp1.is(f31) && !cmp2.is(f31));
+      switch (cc) {
+        case lt:
+          cmp(OLT, L, f31, cmp1, cmp2);
+          bc1nez(target, f31);
+          break;
+        case gt:
+          cmp(ULE, L, f31, cmp1, cmp2);
+          bc1eqz(target, f31);
+          break;
+        case ge:
+          cmp(ULT, L, f31, cmp1, cmp2);
+          bc1eqz(target, f31);
+          break;
+        case le:
+          cmp(OLE, L, f31, cmp1, cmp2);
+          bc1nez(target, f31);
+          break;
+        case eq:
+          cmp(EQ, L, f31, cmp1, cmp2);
+          bc1nez(target, f31);
+          break;
+        case ueq:
+          cmp(UEQ, L, f31, cmp1, cmp2);
+          bc1nez(target, f31);
+          break;
+        case ne:
+          cmp(EQ, L, f31, cmp1, cmp2);
+          bc1eqz(target, f31);
+          break;
+        case nue:
+          cmp(UEQ, L, f31, cmp1, cmp2);
+          bc1eqz(target, f31);
+          break;
+        default:
+          CHECK(0);
+      }
     }
   }
 
@@ -1471,7 +1598,7 @@ void MacroAssembler::Move(FPURegister dst, double imm) {
 
 
 void MacroAssembler::Movz(Register rd, Register rs, Register rt) {
-  if (kArchVariant == kLoongson) {
+  if (kArchVariant == kMips64r6) {
     Label done;
     Branch(&done, ne, rt, Operand(zero_reg));
     mov(rd, rs);
@@ -1483,7 +1610,7 @@ void MacroAssembler::Movz(Register rd, Register rs, Register rt) {
 
 
 void MacroAssembler::Movn(Register rd, Register rs, Register rt) {
-  if (kArchVariant == kLoongson) {
+  if (kArchVariant == kMips64r6) {
     Label done;
     Branch(&done, eq, rt, Operand(zero_reg));
     mov(rd, rs);
@@ -2372,48 +2499,64 @@ void MacroAssembler::BranchAndLinkShort(int16_t offset, Condition cond,
 
       // Signed comparison.
       case greater:
+        // rs > rt
         slt(scratch, r2, rs);
-        daddiu(scratch, scratch, -1);
-        bgezal(scratch, offset);
+        beq(scratch, zero_reg, 2);
+        nop();
+        bal(offset);
         break;
       case greater_equal:
+        // rs >= rt
         slt(scratch, rs, r2);
-        daddiu(scratch, scratch, -1);
-        bltzal(scratch, offset);
+        bne(scratch, zero_reg, 2);
+        nop();
+        bal(offset);
         break;
       case less:
+        // rs < r2
         slt(scratch, rs, r2);
-        daddiu(scratch, scratch, -1);
-        bgezal(scratch, offset);
+        bne(scratch, zero_reg, 2);
+        nop();
+        bal(offset);
         break;
       case less_equal:
+        // rs <= r2
         slt(scratch, r2, rs);
-        daddiu(scratch, scratch, -1);
-        bltzal(scratch, offset);
+        bne(scratch, zero_reg, 2);
+        nop();
+        bal(offset);
         break;
 
+
       // Unsigned comparison.
       case Ugreater:
+        // rs > rt
         sltu(scratch, r2, rs);
-        daddiu(scratch, scratch, -1);
-        bgezal(scratch, offset);
+        beq(scratch, zero_reg, 2);
+        nop();
+        bal(offset);
         break;
       case Ugreater_equal:
+        // rs >= rt
         sltu(scratch, rs, r2);
-        daddiu(scratch, scratch, -1);
-        bltzal(scratch, offset);
+        bne(scratch, zero_reg, 2);
+        nop();
+        bal(offset);
         break;
       case Uless:
+        // rs < r2
         sltu(scratch, rs, r2);
-        daddiu(scratch, scratch, -1);
-        bgezal(scratch, offset);
+        bne(scratch, zero_reg, 2);
+        nop();
+        bal(offset);
         break;
       case Uless_equal:
+        // rs <= r2
         sltu(scratch, r2, rs);
-        daddiu(scratch, scratch, -1);
-        bltzal(scratch, offset);
+        bne(scratch, zero_reg, 2);
+        nop();
+        bal(offset);
         break;
-
       default:
         UNREACHABLE();
     }
@@ -2470,54 +2613,71 @@ void MacroAssembler::BranchAndLinkShort(Label* L, Condition cond, Register rs,
 
       // Signed comparison.
       case greater:
+        // rs > rt
         slt(scratch, r2, rs);
-        daddiu(scratch, scratch, -1);
+        beq(scratch, zero_reg, 2);
+        nop();
         offset = shifted_branch_offset(L, false);
-        bgezal(scratch, offset);
+        bal(offset);
         break;
       case greater_equal:
+        // rs >= rt
         slt(scratch, rs, r2);
-        daddiu(scratch, scratch, -1);
+        bne(scratch, zero_reg, 2);
+        nop();
         offset = shifted_branch_offset(L, false);
-        bltzal(scratch, offset);
+        bal(offset);
         break;
       case less:
+        // rs < r2
         slt(scratch, rs, r2);
-        daddiu(scratch, scratch, -1);
+        bne(scratch, zero_reg, 2);
+        nop();
         offset = shifted_branch_offset(L, false);
-        bgezal(scratch, offset);
+        bal(offset);
         break;
       case less_equal:
+        // rs <= r2
         slt(scratch, r2, rs);
-        daddiu(scratch, scratch, -1);
+        bne(scratch, zero_reg, 2);
+        nop();
         offset = shifted_branch_offset(L, false);
-        bltzal(scratch, offset);
+        bal(offset);
         break;
 
+
       // Unsigned comparison.
       case Ugreater:
+        // rs > rt
         sltu(scratch, r2, rs);
-        daddiu(scratch, scratch, -1);
+        beq(scratch, zero_reg, 2);
+        nop();
         offset = shifted_branch_offset(L, false);
-        bgezal(scratch, offset);
+        bal(offset);
         break;
       case Ugreater_equal:
+        // rs >= rt
         sltu(scratch, rs, r2);
-        daddiu(scratch, scratch, -1);
+        bne(scratch, zero_reg, 2);
+        nop();
         offset = shifted_branch_offset(L, false);
-        bltzal(scratch, offset);
+        bal(offset);
         break;
       case Uless:
+        // rs < r2
         sltu(scratch, rs, r2);
-        daddiu(scratch, scratch, -1);
+        bne(scratch, zero_reg, 2);
+        nop();
         offset = shifted_branch_offset(L, false);
-        bgezal(scratch, offset);
+        bal(offset);
         break;
       case Uless_equal:
+        // rs <= r2
         sltu(scratch, r2, rs);
-        daddiu(scratch, scratch, -1);
+        bne(scratch, zero_reg, 2);
+        nop();
         offset = shifted_branch_offset(L, false);
-        bltzal(scratch, offset);
+        bal(offset);
         break;
 
       default:
@@ -5456,10 +5616,7 @@ void MacroAssembler::CheckPageFlag(
     int mask,
     Condition cc,
     Label* condition_met) {
-  // TODO(plind):  Fix li() so we can use constant embedded inside And().
-  // And(scratch, object, Operand(~Page::kPageAlignmentMask));
-  li(at, Operand(~Page::kPageAlignmentMask), CONSTANT_SIZE);  // plind HACK
-  And(scratch, object, at);
+  And(scratch, object, Operand(~Page::kPageAlignmentMask));
   ld(scratch, MemOperand(scratch, MemoryChunk::kFlagsOffset));
   And(scratch, scratch, Operand(mask));
   Branch(condition_met, cc, scratch, Operand(zero_reg));
@@ -5933,8 +6090,7 @@ void MacroAssembler::TruncatingDiv(Register result,
   ASSERT(!result.is(at));
   MultiplierAndShift ms(divisor);
   li(at, Operand(ms.multiplier()));
-  Mult(dividend, Operand(at));
-  mfhi(result);
+  Mulh(result, dividend, Operand(at));
   if (divisor > 0 && ms.multiplier() < 0) {
     Addu(result, result, Operand(dividend));
   }
diff --git a/src/mips64/macro-assembler-mips64.h b/src/mips64/macro-assembler-mips64.h
index 3ba646343a..d2207624cb 100644
--- a/src/mips64/macro-assembler-mips64.h
+++ b/src/mips64/macro-assembler-mips64.h
@@ -606,10 +606,14 @@ class MacroAssembler: public Assembler {
 
   DEFINE_INSTRUCTION(Addu);
   DEFINE_INSTRUCTION(Daddu);
+  DEFINE_INSTRUCTION(Ddiv);
   DEFINE_INSTRUCTION(Subu);
   DEFINE_INSTRUCTION(Dsubu);
+  DEFINE_INSTRUCTION(Dmod);
   DEFINE_INSTRUCTION(Mul);
+  DEFINE_INSTRUCTION(Mulh);
   DEFINE_INSTRUCTION(Dmul);
+  DEFINE_INSTRUCTION(Dmulh);
   DEFINE_INSTRUCTION2(Mult);
   DEFINE_INSTRUCTION2(Dmult);
   DEFINE_INSTRUCTION2(Multu);
diff --git a/src/mips64/simulator-mips64.cc b/src/mips64/simulator-mips64.cc
index 86e8b91464..f4cad547bb 100644
--- a/src/mips64/simulator-mips64.cc
+++ b/src/mips64/simulator-mips64.cc
@@ -1955,9 +1955,6 @@ void Simulator::ConfigureTypeRegister(Instruction* instr,
   switch (op) {
     case COP1:    // Coprocessor instructions.
       switch (instr->RsFieldRaw()) {
-        case BC1:   // Handled in DecodeTypeImmed, should never come here.
-          UNREACHABLE();
-          break;
         case CFC1:
           // At the moment only FCSR is supported.
           ASSERT(fs_reg == kFCSRRegister);
@@ -1976,8 +1973,6 @@ void Simulator::ConfigureTypeRegister(Instruction* instr,
         case MTC1:
         case DMTC1:
         case MTHC1:
-          // Do the store in the execution step.
-          break;
         case S:
         case D:
         case W:
@@ -1986,7 +1981,8 @@ void Simulator::ConfigureTypeRegister(Instruction* instr,
           // Do everything in the execution step.
           break;
         default:
-          UNIMPLEMENTED_MIPS();
+        // BC1 BC1EQZ BC1NEZ handled in DecodeTypeImmed, should never come here.
+           UNREACHABLE();
       }
       break;
     case COP1X:
@@ -2071,13 +2067,23 @@ void Simulator::ConfigureTypeRegister(Instruction* instr,
         case DSRAV:
           *alu_out = rt >> rs;
           break;
-        case MFHI:
-          *alu_out = get_register(HI);
+        case MFHI:  // MFHI == CLZ on R6.
+          if (kArchVariant != kMips64r6) {
+            ASSERT(instr->SaValue() == 0);
+            *alu_out = get_register(HI);
+          } else {
+            // MIPS spec: If no bits were set in GPR rs, the result written to
+            // GPR rd is 32.
+            // GCC __builtin_clz: If input is 0, the result is undefined.
+            ASSERT(instr->SaValue() == 1);
+            *alu_out =
+                rs_u == 0 ? 32 : CompilerIntrinsics::CountLeadingZeros(rs_u);
+          }
           break;
         case MFLO:
           *alu_out = get_register(LO);
           break;
-        case MULT:
+        case MULT:  // MULT == D_MUL_MUH.
           // TODO(plind) - Unify MULT/DMULT with single set of 64-bit HI/Lo
           // regs.
           // TODO(plind) - make the 32-bit MULT ops conform to spec regarding
@@ -2088,9 +2094,23 @@ void Simulator::ConfigureTypeRegister(Instruction* instr,
         case MULTU:
           *u64hilo = static_cast<uint64_t>(rs_u) * static_cast<uint64_t>(rt_u);
           break;
-        case DMULT:
-          *i128resultH = MultiplyHighSigned(rs, rt);
-          *i128resultL = rs * rt;
+        case DMULT:  // DMULT == D_MUL_MUH.
+          if (kArchVariant != kMips64r6) {
+            *i128resultH = MultiplyHighSigned(rs, rt);
+            *i128resultL = rs * rt;
+          } else {
+            switch (instr->SaValue()) {
+              case MUL_OP:
+                *i128resultL = rs * rt;
+                break;
+              case MUH_OP:
+                *i128resultH = MultiplyHighSigned(rs, rt);
+                break;
+              default:
+                UNIMPLEMENTED_MIPS();
+                break;
+            }
+          }
           break;
         case DMULTU:
           UNIMPLEMENTED_MIPS();
@@ -2295,6 +2315,8 @@ void Simulator::DecodeTypeRegister(Instruction* instr) {
     case COP1:
       switch (instr->RsFieldRaw()) {
         case BC1:   // Branch on coprocessor condition.
+        case BC1EQZ:
+        case BC1NEZ:
           UNREACHABLE();
           break;
         case CFC1:
@@ -2328,20 +2350,9 @@ void Simulator::DecodeTypeRegister(Instruction* instr) {
               f = get_fpu_register_float(fs_reg);
               set_fpu_register_double(fd_reg, static_cast<double>(f));
               break;
-            case CVT_W_S:
-            case CVT_L_S:
-            case TRUNC_W_S:
-            case TRUNC_L_S:
-            case ROUND_W_S:
-            case ROUND_L_S:
-            case FLOOR_W_S:
-            case FLOOR_L_S:
-            case CEIL_W_S:
-            case CEIL_L_S:
-            case CVT_PS_S:
-              UNIMPLEMENTED_MIPS();
-              break;
             default:
+            // CVT_W_S CVT_L_S TRUNC_W_S ROUND_W_S ROUND_L_S FLOOR_W_S FLOOR_L_S
+            // CEIL_W_S CEIL_L_S CVT_PS_S are unimplemented.
               UNREACHABLE();
           }
           break;
@@ -2514,25 +2525,77 @@ void Simulator::DecodeTypeRegister(Instruction* instr) {
               alu_out = get_fpu_register_signed_word(fs_reg);
               set_fpu_register_double(fd_reg, static_cast<double>(alu_out));
               break;
-            default:
+            default:  // Mips64r6 CMP.S instructions unimplemented.
               UNREACHABLE();
           }
           break;
         case L:
+          fs = get_fpu_register_double(fs_reg);
+          ft = get_fpu_register_double(ft_reg);
           switch (instr->FunctionFieldRaw()) {
-          case CVT_D_L:  // Mips32r2 instruction.
-            i64 = get_fpu_register(fs_reg);
-            set_fpu_register_double(fd_reg, static_cast<double>(i64));
-            break;
+            case CVT_D_L:  // Mips32r2 instruction.
+              i64 = get_fpu_register(fs_reg);
+              set_fpu_register_double(fd_reg, static_cast<double>(i64));
+              break;
             case CVT_S_L:
               UNIMPLEMENTED_MIPS();
               break;
-            default:
+            case CMP_AF:  // Mips64r6 CMP.D instructions.
+              UNIMPLEMENTED_MIPS();
+              break;
+            case CMP_UN:
+              if (std::isnan(fs) || std::isnan(ft)) {
+                set_fpu_register(fd_reg, -1);
+              } else {
+                set_fpu_register(fd_reg, 0);
+              }
+              break;
+            case CMP_EQ:
+              if (fs == ft) {
+                set_fpu_register(fd_reg, -1);
+              } else {
+                set_fpu_register(fd_reg, 0);
+              }
+              break;
+            case CMP_UEQ:
+              if ((fs == ft) || (std::isnan(fs) || std::isnan(ft))) {
+                set_fpu_register(fd_reg, -1);
+              } else {
+                set_fpu_register(fd_reg, 0);
+              }
+              break;
+            case CMP_LT:
+              if (fs < ft) {
+                set_fpu_register(fd_reg, -1);
+              } else {
+                set_fpu_register(fd_reg, 0);
+              }
+              break;
+            case CMP_ULT:
+              if ((fs < ft) || (std::isnan(fs) || std::isnan(ft))) {
+                set_fpu_register(fd_reg, -1);
+              } else {
+                set_fpu_register(fd_reg, 0);
+              }
+              break;
+            case CMP_LE:
+              if (fs <= ft) {
+                set_fpu_register(fd_reg, -1);
+              } else {
+                set_fpu_register(fd_reg, 0);
+              }
+              break;
+            case CMP_ULE:
+              if ((fs <= ft) || (std::isnan(fs) || std::isnan(ft))) {
+                set_fpu_register(fd_reg, -1);
+              } else {
+                set_fpu_register(fd_reg, 0);
+              }
+              break;
+            default:  // CMP_OR CMP_UNE CMP_NE UNIMPLEMENTED
               UNREACHABLE();
           }
           break;
-        case PS:
-          break;
         default:
           UNREACHABLE();
       }
@@ -2572,32 +2635,91 @@ void Simulator::DecodeTypeRegister(Instruction* instr) {
         }
         // Instructions using HI and LO registers.
         case MULT:
-          set_register(LO, static_cast<int32_t>(i64hilo & 0xffffffff));
-          set_register(HI, static_cast<int32_t>(i64hilo >> 32));
+          if (kArchVariant != kMips64r6) {
+            set_register(LO, static_cast<int32_t>(i64hilo & 0xffffffff));
+            set_register(HI, static_cast<int32_t>(i64hilo >> 32));
+          } else {
+            switch (instr->SaValue()) {
+              case MUL_OP:
+                set_register(rd_reg,
+                    static_cast<int32_t>(i64hilo & 0xffffffff));
+                break;
+              case MUH_OP:
+                set_register(rd_reg, static_cast<int32_t>(i64hilo >> 32));
+                break;
+              default:
+                UNIMPLEMENTED_MIPS();
+                break;
+            }
+          }
           break;
         case MULTU:
           set_register(LO, static_cast<int32_t>(u64hilo & 0xffffffff));
           set_register(HI, static_cast<int32_t>(u64hilo >> 32));
           break;
-        case DMULT:
-          set_register(LO, static_cast<int64_t>(i128resultL));
-          set_register(HI, static_cast<int64_t>(i128resultH));
+        case DMULT:  // DMULT == D_MUL_MUH.
+          if (kArchVariant != kMips64r6) {
+            set_register(LO, static_cast<int64_t>(i128resultL));
+            set_register(HI, static_cast<int64_t>(i128resultH));
+          } else {
+            switch (instr->SaValue()) {
+              case MUL_OP:
+                set_register(rd_reg, static_cast<int64_t>(i128resultL));
+                break;
+              case MUH_OP:
+                set_register(rd_reg, static_cast<int64_t>(i128resultH));
+                break;
+              default:
+                UNIMPLEMENTED_MIPS();
+                break;
+            }
+          }
           break;
         case DMULTU:
           UNIMPLEMENTED_MIPS();
           break;
+        case DSLL:
+          set_register(rd_reg, alu_out);
+          break;
         case DIV:
         case DDIV:
-          // Divide by zero and overflow was not checked in the configuration
-          // step - div and divu do not raise exceptions. On division by 0
-          // the result will be UNPREDICTABLE. On overflow (INT_MIN/-1),
-          // return INT_MIN which is what the hardware does.
-          if (rs == INT_MIN && rt == -1) {
-            set_register(LO, INT_MIN);
-            set_register(HI, 0);
-          } else if (rt != 0) {
-            set_register(LO, rs / rt);
-            set_register(HI, rs % rt);
+          switch (kArchVariant) {
+            case kMips64r2:
+              // Divide by zero and overflow was not checked in the
+              // configuration step - div and divu do not raise exceptions. On
+              // division by 0 the result will be UNPREDICTABLE. On overflow
+              // (INT_MIN/-1), return INT_MIN which is what the hardware does.
+              if (rs == INT_MIN && rt == -1) {
+                set_register(LO, INT_MIN);
+                set_register(HI, 0);
+              } else if (rt != 0) {
+                set_register(LO, rs / rt);
+                set_register(HI, rs % rt);
+              }
+              break;
+            case kMips64r6:
+              switch (instr->SaValue()) {
+                case DIV_OP:
+                  if (rs == INT_MIN && rt == -1) {
+                    set_register(rd_reg, INT_MIN);
+                  } else if (rt != 0) {
+                    set_register(rd_reg, rs / rt);
+                  }
+                  break;
+                case MOD_OP:
+                  if (rs == INT_MIN && rt == -1) {
+                    set_register(rd_reg, 0);
+                  } else if (rt != 0) {
+                    set_register(rd_reg, rs % rt);
+                  }
+                  break;
+                default:
+                  UNIMPLEMENTED_MIPS();
+                  break;
+              }
+              break;
+            default:
+              break;
           }
           break;
         case DIVU:
@@ -2696,6 +2818,7 @@ void Simulator::DecodeTypeImmediate(Instruction* instr) {
   int16_t  imm16  = instr->Imm16Value();
 
   int32_t  ft_reg = instr->FtValue();  // Destination register.
+  int64_t  ft     = get_fpu_register(ft_reg);
 
   // Zero extended immediate.
   uint32_t  oe_imm16 = 0xffff & imm16;
@@ -2742,6 +2865,26 @@ void Simulator::DecodeTypeImmediate(Instruction* instr) {
             next_pc = current_pc + kBranchReturnOffset;
           }
           break;
+        case BC1EQZ:
+          do_branch = (ft & 0x1) ? false : true;
+          execute_branch_delay_instruction = true;
+          // Set next_pc.
+          if (do_branch) {
+            next_pc = current_pc + (imm16 << 2) + Instruction::kInstrSize;
+          } else {
+            next_pc = current_pc + kBranchReturnOffset;
+          }
+          break;
+        case BC1NEZ:
+          do_branch = (ft & 0x1) ? true : false;
+          execute_branch_delay_instruction = true;
+          // Set next_pc.
+          if (do_branch) {
+            next_pc = current_pc + (imm16 << 2) + Instruction::kInstrSize;
+          } else {
+            next_pc = current_pc + kBranchReturnOffset;
+          }
+          break;
         default:
           UNREACHABLE();
       }
diff --git a/test/cctest/test-api.cc b/test/cctest/test-api.cc
index 95ec479634..d174b307cb 100644
--- a/test/cctest/test-api.cc
+++ b/test/cctest/test-api.cc
@@ -18064,7 +18064,7 @@ THREADED_TEST(QuietSignalingNaNs) {
       uint64_t stored_bits = DoubleToBits(stored_number);
       // Check if quiet nan (bits 51..62 all set).
 #if (defined(V8_TARGET_ARCH_MIPS) || defined(V8_TARGET_ARCH_MIPS64)) && \
-    !defined(USE_SIMULATOR)
+    !defined(_MIPS_ARCH_MIPS64R6) && !defined(USE_SIMULATOR)
       // Most significant fraction bit for quiet nan is set to 0
       // on MIPS architecture. Allowed by IEEE-754.
       CHECK_EQ(0xffe, static_cast<int>((stored_bits >> 51) & 0xfff));
@@ -18085,7 +18085,7 @@ THREADED_TEST(QuietSignalingNaNs) {
       uint64_t stored_bits = DoubleToBits(stored_date);
       // Check if quiet nan (bits 51..62 all set).
 #if (defined(V8_TARGET_ARCH_MIPS) || defined(V8_TARGET_ARCH_MIPS64)) && \
-    !defined(USE_SIMULATOR)
+    !defined(_MIPS_ARCH_MIPS64R6) && !defined(USE_SIMULATOR)
       // Most significant fraction bit for quiet nan is set to 0
       // on MIPS architecture. Allowed by IEEE-754.
       CHECK_EQ(0xffe, static_cast<int>((stored_bits >> 51) & 0xfff));
diff --git a/test/cctest/test-assembler-mips64.cc b/test/cctest/test-assembler-mips64.cc
index bd17595ef1..4e9238930a 100644
--- a/test/cctest/test-assembler-mips64.cc
+++ b/test/cctest/test-assembler-mips64.cc
@@ -557,21 +557,27 @@ TEST(MIPS7) {
 
   __ ldc1(f4, MemOperand(a0, OFFSET_OF(T, a)) );
   __ ldc1(f6, MemOperand(a0, OFFSET_OF(T, b)) );
-  __ c(UN, D, f4, f6);
-  __ bc1f(&neither_is_nan);
+  if (kArchVariant != kMips64r6) {
+    __ c(UN, D, f4, f6);
+    __ bc1f(&neither_is_nan);
+  } else {
+    __ cmp(UN, L, f2, f4, f6);
+    __ bc1eqz(&neither_is_nan, f2);
+  }
   __ nop();
   __ sw(zero_reg, MemOperand(a0, OFFSET_OF(T, result)) );
   __ Branch(&outa_here);
 
   __ bind(&neither_is_nan);
 
-  if (kArchVariant == kLoongson) {
-    __ c(OLT, D, f6, f4);
-    __ bc1t(&less_than);
+  if (kArchVariant == kMips64r6) {
+    __ cmp(OLT, L, f2, f6, f4);
+    __ bc1nez(&less_than, f2);
   } else {
     __ c(OLT, D, f6, f4, 2);
     __ bc1t(&less_than, 2);
   }
+
   __ nop();
   __ sw(zero_reg, MemOperand(a0, OFFSET_OF(T, result)) );
   __ Branch(&outa_here);
@@ -832,144 +838,147 @@ TEST(MIPS10) {
 
 
 TEST(MIPS11) {
-  // Test LWL, LWR, SWL and SWR instructions.
-  CcTest::InitializeVM();
-  Isolate* isolate = CcTest::i_isolate();
-  HandleScope scope(isolate);
+  // Do not run test on MIPS64r6, as these instructions are removed.
+  if (kArchVariant != kMips64r6) {
+    // Test LWL, LWR, SWL and SWR instructions.
+    CcTest::InitializeVM();
+    Isolate* isolate = CcTest::i_isolate();
+    HandleScope scope(isolate);
 
-  typedef struct {
-    int32_t reg_init;
-    int32_t mem_init;
-    int32_t lwl_0;
-    int32_t lwl_1;
-    int32_t lwl_2;
-    int32_t lwl_3;
-    int32_t lwr_0;
-    int32_t lwr_1;
-    int32_t lwr_2;
-    int32_t lwr_3;
-    int32_t swl_0;
-    int32_t swl_1;
-    int32_t swl_2;
-    int32_t swl_3;
-    int32_t swr_0;
-    int32_t swr_1;
-    int32_t swr_2;
-    int32_t swr_3;
-  } T;
-  T t;
+    typedef struct {
+      int32_t reg_init;
+      int32_t mem_init;
+      int32_t lwl_0;
+      int32_t lwl_1;
+      int32_t lwl_2;
+      int32_t lwl_3;
+      int32_t lwr_0;
+      int32_t lwr_1;
+      int32_t lwr_2;
+      int32_t lwr_3;
+      int32_t swl_0;
+      int32_t swl_1;
+      int32_t swl_2;
+      int32_t swl_3;
+      int32_t swr_0;
+      int32_t swr_1;
+      int32_t swr_2;
+      int32_t swr_3;
+    } T;
+    T t;
 
-  Assembler assm(isolate, NULL, 0);
+    Assembler assm(isolate, NULL, 0);
 
-  // Test all combinations of LWL and vAddr.
-  __ lw(a4, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ lwl(a4, MemOperand(a0, OFFSET_OF(T, mem_init)) );
-  __ sw(a4, MemOperand(a0, OFFSET_OF(T, lwl_0)) );
+    // Test all combinations of LWL and vAddr.
+    __ lw(a4, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ lwl(a4, MemOperand(a0, OFFSET_OF(T, mem_init)) );
+    __ sw(a4, MemOperand(a0, OFFSET_OF(T, lwl_0)) );
 
-  __ lw(a5, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ lwl(a5, MemOperand(a0, OFFSET_OF(T, mem_init) + 1) );
-  __ sw(a5, MemOperand(a0, OFFSET_OF(T, lwl_1)) );
+    __ lw(a5, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ lwl(a5, MemOperand(a0, OFFSET_OF(T, mem_init) + 1) );
+    __ sw(a5, MemOperand(a0, OFFSET_OF(T, lwl_1)) );
 
-  __ lw(a6, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ lwl(a6, MemOperand(a0, OFFSET_OF(T, mem_init) + 2) );
-  __ sw(a6, MemOperand(a0, OFFSET_OF(T, lwl_2)) );
+    __ lw(a6, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ lwl(a6, MemOperand(a0, OFFSET_OF(T, mem_init) + 2) );
+    __ sw(a6, MemOperand(a0, OFFSET_OF(T, lwl_2)) );
 
-  __ lw(a7, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ lwl(a7, MemOperand(a0, OFFSET_OF(T, mem_init) + 3) );
-  __ sw(a7, MemOperand(a0, OFFSET_OF(T, lwl_3)) );
+    __ lw(a7, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ lwl(a7, MemOperand(a0, OFFSET_OF(T, mem_init) + 3) );
+    __ sw(a7, MemOperand(a0, OFFSET_OF(T, lwl_3)) );
 
-  // Test all combinations of LWR and vAddr.
-  __ lw(a4, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ lwr(a4, MemOperand(a0, OFFSET_OF(T, mem_init)) );
-  __ sw(a4, MemOperand(a0, OFFSET_OF(T, lwr_0)) );
+    // Test all combinations of LWR and vAddr.
+    __ lw(a4, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ lwr(a4, MemOperand(a0, OFFSET_OF(T, mem_init)) );
+    __ sw(a4, MemOperand(a0, OFFSET_OF(T, lwr_0)) );
 
-  __ lw(a5, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ lwr(a5, MemOperand(a0, OFFSET_OF(T, mem_init) + 1) );
-  __ sw(a5, MemOperand(a0, OFFSET_OF(T, lwr_1)) );
+    __ lw(a5, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ lwr(a5, MemOperand(a0, OFFSET_OF(T, mem_init) + 1) );
+    __ sw(a5, MemOperand(a0, OFFSET_OF(T, lwr_1)) );
 
-  __ lw(a6, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ lwr(a6, MemOperand(a0, OFFSET_OF(T, mem_init) + 2) );
-  __ sw(a6, MemOperand(a0, OFFSET_OF(T, lwr_2)) );
+    __ lw(a6, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ lwr(a6, MemOperand(a0, OFFSET_OF(T, mem_init) + 2) );
+    __ sw(a6, MemOperand(a0, OFFSET_OF(T, lwr_2)) );
 
-  __ lw(a7, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ lwr(a7, MemOperand(a0, OFFSET_OF(T, mem_init) + 3) );
-  __ sw(a7, MemOperand(a0, OFFSET_OF(T, lwr_3)) );
+    __ lw(a7, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ lwr(a7, MemOperand(a0, OFFSET_OF(T, mem_init) + 3) );
+    __ sw(a7, MemOperand(a0, OFFSET_OF(T, lwr_3)) );
 
-  // Test all combinations of SWL and vAddr.
-  __ lw(a4, MemOperand(a0, OFFSET_OF(T, mem_init)) );
-  __ sw(a4, MemOperand(a0, OFFSET_OF(T, swl_0)) );
-  __ lw(a4, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ swl(a4, MemOperand(a0, OFFSET_OF(T, swl_0)) );
+    // Test all combinations of SWL and vAddr.
+    __ lw(a4, MemOperand(a0, OFFSET_OF(T, mem_init)) );
+    __ sw(a4, MemOperand(a0, OFFSET_OF(T, swl_0)) );
+    __ lw(a4, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ swl(a4, MemOperand(a0, OFFSET_OF(T, swl_0)) );
 
-  __ lw(a5, MemOperand(a0, OFFSET_OF(T, mem_init)) );
-  __ sw(a5, MemOperand(a0, OFFSET_OF(T, swl_1)) );
-  __ lw(a5, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ swl(a5, MemOperand(a0, OFFSET_OF(T, swl_1) + 1) );
+    __ lw(a5, MemOperand(a0, OFFSET_OF(T, mem_init)) );
+    __ sw(a5, MemOperand(a0, OFFSET_OF(T, swl_1)) );
+    __ lw(a5, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ swl(a5, MemOperand(a0, OFFSET_OF(T, swl_1) + 1) );
 
-  __ lw(a6, MemOperand(a0, OFFSET_OF(T, mem_init)) );
-  __ sw(a6, MemOperand(a0, OFFSET_OF(T, swl_2)) );
-  __ lw(a6, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ swl(a6, MemOperand(a0, OFFSET_OF(T, swl_2) + 2) );
+    __ lw(a6, MemOperand(a0, OFFSET_OF(T, mem_init)) );
+    __ sw(a6, MemOperand(a0, OFFSET_OF(T, swl_2)) );
+    __ lw(a6, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ swl(a6, MemOperand(a0, OFFSET_OF(T, swl_2) + 2) );
 
-  __ lw(a7, MemOperand(a0, OFFSET_OF(T, mem_init)) );
-  __ sw(a7, MemOperand(a0, OFFSET_OF(T, swl_3)) );
-  __ lw(a7, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ swl(a7, MemOperand(a0, OFFSET_OF(T, swl_3) + 3) );
+    __ lw(a7, MemOperand(a0, OFFSET_OF(T, mem_init)) );
+    __ sw(a7, MemOperand(a0, OFFSET_OF(T, swl_3)) );
+    __ lw(a7, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ swl(a7, MemOperand(a0, OFFSET_OF(T, swl_3) + 3) );
 
-  // Test all combinations of SWR and vAddr.
-  __ lw(a4, MemOperand(a0, OFFSET_OF(T, mem_init)) );
-  __ sw(a4, MemOperand(a0, OFFSET_OF(T, swr_0)) );
-  __ lw(a4, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ swr(a4, MemOperand(a0, OFFSET_OF(T, swr_0)) );
+    // Test all combinations of SWR and vAddr.
+    __ lw(a4, MemOperand(a0, OFFSET_OF(T, mem_init)) );
+    __ sw(a4, MemOperand(a0, OFFSET_OF(T, swr_0)) );
+    __ lw(a4, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ swr(a4, MemOperand(a0, OFFSET_OF(T, swr_0)) );
 
-  __ lw(a5, MemOperand(a0, OFFSET_OF(T, mem_init)) );
-  __ sw(a5, MemOperand(a0, OFFSET_OF(T, swr_1)) );
-  __ lw(a5, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ swr(a5, MemOperand(a0, OFFSET_OF(T, swr_1) + 1) );
+    __ lw(a5, MemOperand(a0, OFFSET_OF(T, mem_init)) );
+    __ sw(a5, MemOperand(a0, OFFSET_OF(T, swr_1)) );
+    __ lw(a5, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ swr(a5, MemOperand(a0, OFFSET_OF(T, swr_1) + 1) );
 
-  __ lw(a6, MemOperand(a0, OFFSET_OF(T, mem_init)) );
-  __ sw(a6, MemOperand(a0, OFFSET_OF(T, swr_2)) );
-  __ lw(a6, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ swr(a6, MemOperand(a0, OFFSET_OF(T, swr_2) + 2) );
+    __ lw(a6, MemOperand(a0, OFFSET_OF(T, mem_init)) );
+    __ sw(a6, MemOperand(a0, OFFSET_OF(T, swr_2)) );
+    __ lw(a6, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ swr(a6, MemOperand(a0, OFFSET_OF(T, swr_2) + 2) );
 
-  __ lw(a7, MemOperand(a0, OFFSET_OF(T, mem_init)) );
-  __ sw(a7, MemOperand(a0, OFFSET_OF(T, swr_3)) );
-  __ lw(a7, MemOperand(a0, OFFSET_OF(T, reg_init)) );
-  __ swr(a7, MemOperand(a0, OFFSET_OF(T, swr_3) + 3) );
+    __ lw(a7, MemOperand(a0, OFFSET_OF(T, mem_init)) );
+    __ sw(a7, MemOperand(a0, OFFSET_OF(T, swr_3)) );
+    __ lw(a7, MemOperand(a0, OFFSET_OF(T, reg_init)) );
+    __ swr(a7, MemOperand(a0, OFFSET_OF(T, swr_3) + 3) );
 
-  __ jr(ra);
-  __ nop();
+    __ jr(ra);
+    __ nop();
 
-  CodeDesc desc;
-  assm.GetCode(&desc);
-  Handle<Code> code = isolate->factory()->NewCode(
-      desc, Code::ComputeFlags(Code::STUB), Handle<Code>());
-  F3 f = FUNCTION_CAST<F3>(code->entry());
-  t.reg_init = 0xaabbccdd;
-  t.mem_init = 0x11223344;
+    CodeDesc desc;
+    assm.GetCode(&desc);
+    Handle<Code> code = isolate->factory()->NewCode(
+        desc, Code::ComputeFlags(Code::STUB), Handle<Code>());
+    F3 f = FUNCTION_CAST<F3>(code->entry());
+    t.reg_init = 0xaabbccdd;
+    t.mem_init = 0x11223344;
 
-  Object* dummy = CALL_GENERATED_CODE(f, &t, 0, 0, 0, 0);
-  USE(dummy);
+    Object* dummy = CALL_GENERATED_CODE(f, &t, 0, 0, 0, 0);
+    USE(dummy);
 
-  CHECK_EQ(0x44bbccdd, t.lwl_0);
-  CHECK_EQ(0x3344ccdd, t.lwl_1);
-  CHECK_EQ(0x223344dd, t.lwl_2);
-  CHECK_EQ(0x11223344, t.lwl_3);
+    CHECK_EQ(0x44bbccdd, t.lwl_0);
+    CHECK_EQ(0x3344ccdd, t.lwl_1);
+    CHECK_EQ(0x223344dd, t.lwl_2);
+    CHECK_EQ(0x11223344, t.lwl_3);
 
-  CHECK_EQ(0x11223344, t.lwr_0);
-  CHECK_EQ(0xaa112233, t.lwr_1);
-  CHECK_EQ(0xaabb1122, t.lwr_2);
-  CHECK_EQ(0xaabbcc11, t.lwr_3);
+    CHECK_EQ(0x11223344, t.lwr_0);
+    CHECK_EQ(0xaa112233, t.lwr_1);
+    CHECK_EQ(0xaabb1122, t.lwr_2);
+    CHECK_EQ(0xaabbcc11, t.lwr_3);
 
-  CHECK_EQ(0x112233aa, t.swl_0);
-  CHECK_EQ(0x1122aabb, t.swl_1);
-  CHECK_EQ(0x11aabbcc, t.swl_2);
-  CHECK_EQ(0xaabbccdd, t.swl_3);
+    CHECK_EQ(0x112233aa, t.swl_0);
+    CHECK_EQ(0x1122aabb, t.swl_1);
+    CHECK_EQ(0x11aabbcc, t.swl_2);
+    CHECK_EQ(0xaabbccdd, t.swl_3);
 
-  CHECK_EQ(0xaabbccdd, t.swr_0);
-  CHECK_EQ(0xbbccdd44, t.swr_1);
-  CHECK_EQ(0xccdd3344, t.swr_2);
-  CHECK_EQ(0xdd223344, t.swr_3);
+    CHECK_EQ(0xaabbccdd, t.swr_0);
+    CHECK_EQ(0xbbccdd44, t.swr_1);
+    CHECK_EQ(0xccdd3344, t.swr_2);
+    CHECK_EQ(0xdd223344, t.swr_3);
+  }
 }
 
 
diff --git a/test/cctest/test-disasm-mips64.cc b/test/cctest/test-disasm-mips64.cc
index c615afbf46..d682d33480 100644
--- a/test/cctest/test-disasm-mips64.cc
+++ b/test/cctest/test-disasm-mips64.cc
@@ -122,65 +122,196 @@ TEST(Type0) {
   COMPARE(dsubu(v0, v1, s0),
           "0070102f       dsubu   v0, v1, s0");
 
-  COMPARE(mult(a0, a1),
-          "00850018       mult    a0, a1");
-  COMPARE(dmult(a0, a1),
-          "0085001c       dmult   a0, a1");
-  COMPARE(mult(a6, a7),
-          "014b0018       mult    a6, a7");
-  COMPARE(dmult(a6, a7),
-          "014b001c       dmult   a6, a7");
-  COMPARE(mult(v0, v1),
-          "00430018       mult    v0, v1");
-  COMPARE(dmult(v0, v1),
-          "0043001c       dmult   v0, v1");
+  if (kArchVariant != kMips64r6) {
+    COMPARE(mult(a0, a1),
+            "00850018       mult    a0, a1");
+    COMPARE(dmult(a0, a1),
+            "0085001c       dmult   a0, a1");
+    COMPARE(mult(a6, a7),
+            "014b0018       mult    a6, a7");
+    COMPARE(dmult(a6, a7),
+            "014b001c       dmult   a6, a7");
+    COMPARE(mult(v0, v1),
+            "00430018       mult    v0, v1");
+    COMPARE(dmult(v0, v1),
+            "0043001c       dmult   v0, v1");
 
-  COMPARE(multu(a0, a1),
-          "00850019       multu   a0, a1");
-  COMPARE(dmultu(a0, a1),
-          "0085001d       dmultu  a0, a1");
-  COMPARE(multu(a6, a7),
-          "014b0019       multu   a6, a7");
-  COMPARE(dmultu(a6, a7),
-          "014b001d       dmultu  a6, a7");
-  COMPARE(multu(v0, v1),
-          "00430019       multu   v0, v1");
-  COMPARE(dmultu(v0, v1),
-          "0043001d       dmultu  v0, v1");
+    COMPARE(multu(a0, a1),
+            "00850019       multu   a0, a1");
+    COMPARE(dmultu(a0, a1),
+            "0085001d       dmultu  a0, a1");
+    COMPARE(multu(a6, a7),
+            "014b0019       multu   a6, a7");
+    COMPARE(dmultu(a6, a7),
+            "014b001d       dmultu  a6, a7");
+    COMPARE(multu(v0, v1),
+            "00430019       multu   v0, v1");
+    COMPARE(dmultu(v0, v1),
+            "0043001d       dmultu  v0, v1");
 
-  COMPARE(div(a0, a1),
-          "0085001a       div     a0, a1");
-  COMPARE(div(a6, a7),
-          "014b001a       div     a6, a7");
-  COMPARE(div(v0, v1),
-          "0043001a       div     v0, v1");
-  COMPARE(ddiv(a0, a1),
-          "0085001e       ddiv    a0, a1");
-  COMPARE(ddiv(a6, a7),
-          "014b001e       ddiv    a6, a7");
-  COMPARE(ddiv(v0, v1),
-          "0043001e       ddiv    v0, v1");
+    COMPARE(div(a0, a1),
+            "0085001a       div     a0, a1");
+    COMPARE(div(a6, a7),
+            "014b001a       div     a6, a7");
+    COMPARE(div(v0, v1),
+            "0043001a       div     v0, v1");
+    COMPARE(ddiv(a0, a1),
+            "0085001e       ddiv    a0, a1");
+    COMPARE(ddiv(a6, a7),
+            "014b001e       ddiv    a6, a7");
+    COMPARE(ddiv(v0, v1),
+            "0043001e       ddiv    v0, v1");
 
-  COMPARE(divu(a0, a1),
-          "0085001b       divu    a0, a1");
-  COMPARE(divu(a6, a7),
-          "014b001b       divu    a6, a7");
-  COMPARE(divu(v0, v1),
-          "0043001b       divu    v0, v1");
-  COMPARE(ddivu(a0, a1),
-          "0085001f       ddivu   a0, a1");
-  COMPARE(ddivu(a6, a7),
-          "014b001f       ddivu   a6, a7");
-  COMPARE(ddivu(v0, v1),
-          "0043001f       ddivu   v0, v1");
-
-  if (kArchVariant != kLoongson) {
+    COMPARE(divu(a0, a1),
+            "0085001b       divu    a0, a1");
+    COMPARE(divu(a6, a7),
+            "014b001b       divu    a6, a7");
+    COMPARE(divu(v0, v1),
+            "0043001b       divu    v0, v1");
+    COMPARE(ddivu(a0, a1),
+            "0085001f       ddivu   a0, a1");
+    COMPARE(ddivu(a6, a7),
+            "014b001f       ddivu   a6, a7");
+    COMPARE(ddivu(v0, v1),
+            "0043001f       ddivu   v0, v1");
     COMPARE(mul(a0, a1, a2),
             "70a62002       mul     a0, a1, a2");
     COMPARE(mul(a6, a7, t0),
             "716c5002       mul     a6, a7, t0");
     COMPARE(mul(v0, v1, s0),
             "70701002       mul     v0, v1, s0");
+  } else {  // MIPS64r6.
+    COMPARE(mul(a0, a1, a2),
+            "00a62098       mul    a0, a1, a2");
+    COMPARE(muh(a0, a1, a2),
+            "00a620d8       muh    a0, a1, a2");
+    COMPARE(dmul(a0, a1, a2),
+            "00a6209c       dmul   a0, a1, a2");
+    COMPARE(dmuh(a0, a1, a2),
+            "00a620dc       dmuh   a0, a1, a2");
+    COMPARE(mul(a5, a6, a7),
+            "014b4898       mul    a5, a6, a7");
+    COMPARE(muh(a5, a6, a7),
+            "014b48d8       muh    a5, a6, a7");
+    COMPARE(dmul(a5, a6, a7),
+            "014b489c       dmul   a5, a6, a7");
+    COMPARE(dmuh(a5, a6, a7),
+            "014b48dc       dmuh   a5, a6, a7");
+    COMPARE(mul(v0, v1, a0),
+            "00641098       mul    v0, v1, a0");
+    COMPARE(muh(v0, v1, a0),
+            "006410d8       muh    v0, v1, a0");
+    COMPARE(dmul(v0, v1, a0),
+            "0064109c       dmul   v0, v1, a0");
+    COMPARE(dmuh(v0, v1, a0),
+            "006410dc       dmuh   v0, v1, a0");
+
+    COMPARE(mulu(a0, a1, a2),
+            "00a62099       mulu   a0, a1, a2");
+    COMPARE(muhu(a0, a1, a2),
+            "00a620d9       muhu   a0, a1, a2");
+    COMPARE(dmulu(a0, a1, a2),
+            "00a6209d       dmulu  a0, a1, a2");
+    COMPARE(dmuhu(a0, a1, a2),
+            "00a620dd       dmuhu  a0, a1, a2");
+    COMPARE(mulu(a5, a6, a7),
+            "014b4899       mulu   a5, a6, a7");
+    COMPARE(muhu(a5, a6, a7),
+            "014b48d9       muhu   a5, a6, a7");
+    COMPARE(dmulu(a5, a6, a7),
+            "014b489d       dmulu  a5, a6, a7");
+    COMPARE(dmuhu(a5, a6, a7),
+            "014b48dd       dmuhu  a5, a6, a7");
+    COMPARE(mulu(v0, v1, a0),
+            "00641099       mulu   v0, v1, a0");
+    COMPARE(muhu(v0, v1, a0),
+            "006410d9       muhu   v0, v1, a0");
+    COMPARE(dmulu(v0, v1, a0),
+            "0064109d       dmulu  v0, v1, a0");
+    COMPARE(dmuhu(v0, v1, a0),
+            "006410dd       dmuhu  v0, v1, a0");
+
+    COMPARE(div(a0, a1, a2),
+            "00a6209a       div    a0, a1, a2");
+    COMPARE(mod(a0, a1, a2),
+            "00a620da       mod    a0, a1, a2");
+    COMPARE(ddiv(a0, a1, a2),
+            "00a6209e       ddiv   a0, a1, a2");
+    COMPARE(dmod(a0, a1, a2),
+            "00a620de       dmod   a0, a1, a2");
+    COMPARE(div(a5, a6, a7),
+            "014b489a       div    a5, a6, a7");
+    COMPARE(mod(a5, a6, a7),
+            "014b48da       mod    a5, a6, a7");
+    COMPARE(ddiv(a5, a6, a7),
+            "014b489e       ddiv   a5, a6, a7");
+    COMPARE(dmod(a5, a6, a7),
+            "014b48de       dmod   a5, a6, a7");
+    COMPARE(div(v0, v1, a0),
+            "0064109a       div    v0, v1, a0");
+    COMPARE(mod(v0, v1, a0),
+            "006410da       mod    v0, v1, a0");
+    COMPARE(ddiv(v0, v1, a0),
+            "0064109e       ddiv   v0, v1, a0");
+    COMPARE(dmod(v0, v1, a0),
+            "006410de       dmod   v0, v1, a0");
+
+    COMPARE(divu(a0, a1, a2),
+            "00a6209b       divu   a0, a1, a2");
+    COMPARE(modu(a0, a1, a2),
+            "00a620db       modu   a0, a1, a2");
+    COMPARE(ddivu(a0, a1, a2),
+            "00a6209f       ddivu  a0, a1, a2");
+    COMPARE(dmodu(a0, a1, a2),
+            "00a620df       dmodu  a0, a1, a2");
+    COMPARE(divu(a5, a6, a7),
+            "014b489b       divu   a5, a6, a7");
+    COMPARE(modu(a5, a6, a7),
+            "014b48db       modu   a5, a6, a7");
+    COMPARE(ddivu(a5, a6, a7),
+            "014b489f       ddivu  a5, a6, a7");
+    COMPARE(dmodu(a5, a6, a7),
+            "014b48df       dmodu  a5, a6, a7");
+    COMPARE(divu(v0, v1, a0),
+            "0064109b       divu   v0, v1, a0");
+    COMPARE(modu(v0, v1, a0),
+            "006410db       modu   v0, v1, a0");
+    COMPARE(ddivu(v0, v1, a0),
+            "0064109f       ddivu  v0, v1, a0");
+    COMPARE(dmodu(v0, v1, a0),
+            "006410df       dmodu  v0, v1, a0");
+
+    COMPARE(bovc(a0, a0, static_cast<int16_t>(0)),
+            "20840000       bovc  a0, a0, 0");
+    COMPARE(bovc(a1, a0, static_cast<int16_t>(0)),
+            "20a40000       bovc  a1, a0, 0");
+    COMPARE(bovc(a1, a0, 32767),
+            "20a47fff       bovc  a1, a0, 32767");
+    COMPARE(bovc(a1, a0, -32768),
+            "20a48000       bovc  a1, a0, -32768");
+
+    COMPARE(bnvc(a0, a0, static_cast<int16_t>(0)),
+            "60840000       bnvc  a0, a0, 0");
+    COMPARE(bnvc(a1, a0, static_cast<int16_t>(0)),
+            "60a40000       bnvc  a1, a0, 0");
+    COMPARE(bnvc(a1, a0, 32767),
+            "60a47fff       bnvc  a1, a0, 32767");
+    COMPARE(bnvc(a1, a0, -32768),
+            "60a48000       bnvc  a1, a0, -32768");
+
+    COMPARE(beqzc(a0, 0),
+            "d8800000       beqzc   a0, 0x0");
+    COMPARE(beqzc(a0, 0xfffff),                   // 0x0fffff ==  1048575.
+            "d88fffff       beqzc   a0, 0xfffff");
+    COMPARE(beqzc(a0, 0x100000),                  // 0x100000 == -1048576.
+            "d8900000       beqzc   a0, 0x100000");
+
+    COMPARE(bnezc(a0, 0),
+            "f8800000       bnezc   a0, 0x0");
+    COMPARE(bnezc(a0, 0xfffff),                   // 0x0fffff ==  1048575.
+            "f88fffff       bnezc   a0, 0xfffff");
+    COMPARE(bnezc(a0, 0x100000),                  // 0x100000 == -1048576.
+            "f8900000       bnezc   a0, 0x100000");
   }
 
   COMPARE(addiu(a0, a1, 0x0),
@@ -476,42 +607,48 @@ TEST(Type0) {
           "2d6a8000       sltiu   a6, a7, -32768");
   COMPARE(sltiu(v0, v1, -1),
           "2c62ffff       sltiu   v0, v1, -1");
+  COMPARE(movz(a0, a1, a2),
+          "00a6200a       movz    a0, a1, a2");
+  COMPARE(movz(s0, s1, s2),
+          "0232800a       movz    s0, s1, s2");
+  COMPARE(movz(a6, a7, t0),
+          "016c500a       movz    a6, a7, t0");
+  COMPARE(movz(v0, v1, a2),
+          "0066100a       movz    v0, v1, a2");
+  COMPARE(movn(a0, a1, a2),
+          "00a6200b       movn    a0, a1, a2");
+  COMPARE(movn(s0, s1, s2),
+          "0232800b       movn    s0, s1, s2");
+  COMPARE(movn(a6, a7, t0),
+          "016c500b       movn    a6, a7, t0");
+  COMPARE(movn(v0, v1, a2),
+          "0066100b       movn    v0, v1, a2");
 
-  if (kArchVariant != kLoongson) {
-    COMPARE(movz(a0, a1, a2),
-            "00a6200a       movz    a0, a1, a2");
-    COMPARE(movz(s0, s1, s2),
-            "0232800a       movz    s0, s1, s2");
-    COMPARE(movz(a6, a7, t0),
-            "016c500a       movz    a6, a7, t0");
-    COMPARE(movz(v0, v1, a2),
-            "0066100a       movz    v0, v1, a2");
-    COMPARE(movn(a0, a1, a2),
-            "00a6200b       movn    a0, a1, a2");
-    COMPARE(movn(s0, s1, s2),
-            "0232800b       movn    s0, s1, s2");
-    COMPARE(movn(a6, a7, t0),
-            "016c500b       movn    a6, a7, t0");
-    COMPARE(movn(v0, v1, a2),
-            "0066100b       movn    v0, v1, a2");
-
-    COMPARE(movt(a0, a1, 1),
-            "00a52001       movt    a0, a1, 1");
-    COMPARE(movt(s0, s1, 2),
-            "02298001       movt    s0, s1, 2");
-    COMPARE(movt(a6, a7, 3),
-            "016d5001       movt    a6, a7, 3");
-    COMPARE(movt(v0, v1, 7),
-            "007d1001       movt    v0, v1, 7");
-    COMPARE(movf(a0, a1, 0),
-            "00a02001       movf    a0, a1, 0");
-    COMPARE(movf(s0, s1, 4),
-            "02308001       movf    s0, s1, 4");
-    COMPARE(movf(a6, a7, 5),
-            "01745001       movf    a6, a7, 5");
-    COMPARE(movf(v0, v1, 6),
-            "00781001       movf    v0, v1, 6");
+  COMPARE(movt(a0, a1, 1),
+          "00a52001       movt    a0, a1, 1");
+  COMPARE(movt(s0, s1, 2),
+          "02298001       movt    s0, s1, 2");
+  COMPARE(movt(a6, a7, 3),
+          "016d5001       movt    a6, a7, 3");
+  COMPARE(movt(v0, v1, 7),
+          "007d1001       movt    v0, v1, 7");
+  COMPARE(movf(a0, a1, 0),
+          "00a02001       movf    a0, a1, 0");
+  COMPARE(movf(s0, s1, 4),
+          "02308001       movf    s0, s1, 4");
+  COMPARE(movf(a6, a7, 5),
+          "01745001       movf    a6, a7, 5");
+  COMPARE(movf(v0, v1, 6),
+          "00781001       movf    v0, v1, 6");
 
+  if (kArchVariant == kMips64r6) {
+    COMPARE(clz(a0, a1),
+            "00a02050       clz     a0, a1");
+    COMPARE(clz(s6, s7),
+            "02e0b050       clz     s6, s7");
+    COMPARE(clz(v0, v1),
+            "00601050       clz     v0, v1");
+  } else {
     COMPARE(clz(a0, a1),
             "70a42020       clz     a0, a1");
     COMPARE(clz(s6, s7),
@@ -520,20 +657,18 @@ TEST(Type0) {
             "70621020       clz     v0, v1");
   }
 
-  if (kArchVariant == kMips64r2) {
-    COMPARE(ins_(a0, a1, 31, 1),
-            "7ca4ffc4       ins     a0, a1, 31, 1");
-    COMPARE(ins_(s6, s7, 30, 2),
-            "7ef6ff84       ins     s6, s7, 30, 2");
-    COMPARE(ins_(v0, v1, 0, 32),
-            "7c62f804       ins     v0, v1, 0, 32");
-    COMPARE(ext_(a0, a1, 31, 1),
-            "7ca407c0       ext     a0, a1, 31, 1");
-    COMPARE(ext_(s6, s7, 30, 2),
-            "7ef60f80       ext     s6, s7, 30, 2");
-    COMPARE(ext_(v0, v1, 0, 32),
-            "7c62f800       ext     v0, v1, 0, 32");
-  }
+  COMPARE(ins_(a0, a1, 31, 1),
+          "7ca4ffc4       ins     a0, a1, 31, 1");
+  COMPARE(ins_(s6, s7, 30, 2),
+          "7ef6ff84       ins     s6, s7, 30, 2");
+  COMPARE(ins_(v0, v1, 0, 32),
+          "7c62f804       ins     v0, v1, 0, 32");
+  COMPARE(ext_(a0, a1, 31, 1),
+          "7ca407c0       ext     a0, a1, 31, 1");
+  COMPARE(ext_(s6, s7, 30, 2),
+          "7ef60f80       ext     s6, s7, 30, 2");
+  COMPARE(ext_(v0, v1, 0, 32),
+          "7c62f800       ext     v0, v1, 0, 32");
 
   VERIFY_RUN();
 }
diff --git a/test/mjsunit/mjsunit.status b/test/mjsunit/mjsunit.status
index e3de741dea..f329deb360 100644
--- a/test/mjsunit/mjsunit.status
+++ b/test/mjsunit/mjsunit.status
@@ -414,6 +414,9 @@
 
   # Currently always deopt on minus zero
   'math-floor-of-div-minus-zero': [SKIP],
+
+  # BUG(v8:3457).
+  'deserialize-reference': [SKIP],
 }],  # 'arch == mips64el'
 
 ['arch == mips64el and simulator_run == False', {